High Performance Modular Wood Construction Systems · Project No. UNB5 Value to Wood Project...

Project No. UNB5 Value to Wood Project Research Report 2007

High Performance Modular Wood Construction Systems

by

Ian Smith, Andi Asiz, and Gopinath Gupta

Wood Science and Technology Centre Faculty of Forestry and Environmental Management

University of New Brunswick, Fredericton

This report was produced as part of the Value to Wood Program, funded by Natural Resources Canada

April 2007

UNB5 High Performance Modular Wood Construction Systems of 151

Summary Prefabricated (also known as prefab or modular) wood construction is an important part of the Canadian building industry. In 2004 domestic sales of prefab housing units were valued at $700 million, in addition to export sales of $197 million. Prefab housing production in North America as a whole is about 300,000 units per year and valued at $11 billion USD. Consumer acceptance of such products is rising steadily and emphasis is turning towards ‘higher end’ designs. Despite this, the prefab housing industry lags behind other sectors in respect of using advanced technological methods of design and manufacturing. Prefab construction practices in North America are quite labour intensive with some manufacturing steps having simply been taken from the field (as in traditional ‘stick-built’ wood construction’) into the factory setting. Processes have not been fully reengineered and the industry needs to increase its efficiency by adopting innovative design, manufacturing, shipping and site technologies. The goal of this project was to create and make available to industry in Canada the best possible know-how needed to improve the quality and durability, and marketability of prefab wood buildings. Based on analysis of the status of the industry and discussion with project advisors, the project emphasised: the building physics performance of buildings constructed with prefab wall panels, and the structural performance of prefab mini homes during post assembly line handling and road transportation. Knowledge about both these topics is essential for improving the quality and lifespan of completed buildings. In the building physics aspect focus was on investigating air leakage and moisture deposition (hygrothermal) performance of wood walls, and especially construction joints in panellized prefab light-frame systems. Construction joints are unavoidable features of panellized construction and can be the source of many in-service performance problems. Apart from structural integrity requirements, joints must be air tight to avoid leakage of heating/cooling energy and prevent moisture depositions within wall materials. The research involved field monitoring of a building in Fredericton, NB that has prefab wall panels, and development of numerical computer models of heat and moisture transfer in such walls. During the field monitoring hygrothermal parameters like temperature, relative humidity and differential pressures across layers of materials were recorded. Data that was collected is unique in that it articulated the behaviour of ‘butt’ type construction joints in walls, and not just wall panels as in other studies. Results showed that air movement occurred mostly in the construction joints compared to the clear (joint free) wall sections. Air leakage helped dry materials around construction joints but was wasteful of energy. No excessive wetting was observed anywhere in walls. It was concluded that the current ‘typical’ butt method of making construction joints should be improved. To investigate panellized joint configurations other than butt-joints, numerical models of heat and moisture transfer were developed using finite element methods. The modelling is unique in terms of the numerical scheme. Previously scientists had employed finite difference/volume schemes to solve heat and moisture transfer in building assemblies. Such models are adequate analysis of simple situations where walls are of uniform and homogenous composition. However, real buildings contain non-regular geometry; walls contain joints and gaps, and are not uniform. Only finite element schemes have the flexibility needed to analyse realistic problems.


The model created was verified using the field monitoring hygrothermal data for the building in Fredericton with panellized walls. Verification indicated that the model is accurate. It was used to study how the leakage performance of construction joints can be improved without creating unfavourable moisture situations that compromise durability of prefab buildings. Structural performance issues for prefab modular buildings and mini homes concern avoidance of damage during post assembly line handling and road transportation. The problems are not new and have been studied before, but solutions have not been adequate. Modules and mini homes must be designed with sufficient resistance to static and dynamic stresses arising during handling, transportation and site installation. It is reported that in the attempt to avoid handling and transportation damage, manufacturers of mini homes use up to 30% more lumber than is required by building codes that apply to site-built homes. However, adding extra material can actually exacerbate rather than ameliorate problems, and can increase the extent of damage. Why is because extra materials are placed without the aid of the proper know-how and without scientific study of the ‘causes and effects’. Most typical damages are cracking of plasterboard in walls and ceilings. This does not usually impact the in-service structural performance significantly, but damage can compromise heating efficiency and cause moisture problems that lead to rot and mould growth. Public perception that prefab homes are low quality homes is often because of the minor damages that occur during handling and transportation. Research activities involved field observation of a prefab light-frame mini home during factory handling and road transportation, and development of numerical structural models. In field tests parameters measured included deformations, vibration forces and wind pressures. No tests of similar complexity had been conducted before. Data indicated that damages were initiated mostly during factory lifting rather than road transportation. Particularly vulnerable locations are plasterboard near corners of window and doors and at wall to ceiling junctions. The main problems observed concerned the inadequate number and placement of lifting jacks during factory yard handling, and unevenness of jacking processes. It should be cautioned that although these were the major problems for the study situation, the specifics findings might not be generally valid. Analytical models are required for generalised study of handling and transportation problems with prefab wood construction. Numerical models of prefab housing units were created to address efficiency of the current employed structural reinforcing techniques, and as tools to identify best strategies for mitigating handling and transportation damages. Models were three-dimensional and based on static and dynamic finite element analysis techniques. By applying the verified models to case studies it is demonstrated that adding extra lifting supports greatly reduces stresses at the crucial locations during lifting and therefore mitigates the potential for causing damage. Also, it has been shown that materials efficiencies are possible in terms of reducing the amount of lumber framing materials. With proper design of process practices and the structural system it should be possible to eliminate damage and reduce manufacturing costs. However, those solutions like solution of hygrothermal problems must be tailored to specific products and how they will be used/treated once they leave the factory gate.


Acknowledgements The University of New Brunswick acknowledges the financial support of this project by Natural Resources Canada. Thanks are due to the project champion Mr. Kenneth Koo of Jager Building Systems and industry liaisons Mr. Ruzz Muzyka of Nu-Fab Building Products Ltd, and Mr. Hank Starno of the Manufactured Housing Institute. Their support and technical advice is much appreciated. The project team also thanks Jager Building Systems, Ontario for providing LVL materials.

Research Staff

• Professor (Dr.) Ian Smith, Project Leader. • Dr. Andi Asiz, Research Associate. • Mr. Gopinath Gupta, Graduate Student Research Assistant. • Dr. Chuanshuang Hu, Post-Doctoral Fellow. • Mr. Donghua Jia, Graduate Student Assistant. • Mr. Amir Aghakanlou, Graduate Student Assistant. • Mr. Dean McCarthy, Research Technician.


Table of Contents

Summary......................................................................................................................... 2 Acknowledgements ......................................................................................................... 4 Research Staff ................................................................................................................ 4 Table of Contents............................................................................................................ 5 List of Tables................................................................................................................... 7 List of Figures.................................................................................................................. 8 1. Goal and project objectives ......................................................................................... 9 2. Introduction ................................................................................................................. 9 3. Overview of method .................................................................................................. 12 PART I........................................................................................................................... 14 4. Building physics performance aspect of panellized wood buildings .......................... 15

4.1 Background ......................................................................................................... 15 4.2 Long-term field monitoring program..................................................................... 16

4.2.1 Related field-monitoring activities.................................................................. 16 4.2.2 Test building ............................................................................................ 17 4.2.3 Test wall................................................................................................... 19 4.2.4 Field sensors and other instrumentation.................................................. 20 4.2.5 Monitoring and recording ......................................................................... 23 4.2.6 Field test results....................................................................................... 23 4.2.7 Summary of the field monitoring results................................................... 30

4.3 Numerical modelling of heat and moisture transfer.............................................. 31 4.3.1 Governing equations ..................................................................................... 32 4.3.2 Domain and boundary conditions.................................................................. 33 4.3.3 Material properties ........................................................................................ 35 4.3.4 Numerical solution procedure........................................................................ 38 4.3.5 Model verification .......................................................................................... 38 4.3.6 Prefab wall panel-to-wall panel joint analysis ................................................ 44

PART II.......................................................................................................................... 47 5. Handling and transportation aspects of modular wood buildings .............................. 48

5.1 Background ......................................................................................................... 48 5.2 Related Work....................................................................................................... 49 5.3 Field test program................................................................................................ 50

5.3.1 Parameters measured................................................................................... 50 5.3.2 Description of the prefab home tested .......................................................... 50 5.3.3 Field instrumentation..................................................................................... 51 5.3.4 Field test procedure ...................................................................................... 54 5.3.5 Field test results ............................................................................................ 55

5.4 Numerical modelling ............................................................................................ 59


5.4.1 Finite element model ........................................................................................ 59 5.4.2 Boundary conditions (B.C.) ........................................................................... 60 5.4.3 Materials and joints/connections models....................................................... 63 5.4.4 Loads ............................................................................................................ 65 5.4.5 Numerical modelling results .......................................................................... 65 5.4.6 Stress analysis .............................................................................................. 67 5.4.7 Sensitivity analysis ........................................................................................ 70

5.5 Summary of findings for handling and transportation........................................... 73 6. Conclusions, recommendations and impacts ............................................................ 81

6.1 Conclusions ......................................................................................................... 81 6.2 Recommendations for future work ....................................................................... 82 6.3 Impact/benefits to the prefab wood industry ........................................................ 83

7. References................................................................................................................ 84 Appendix A - Temperatures Appendix B - Relative Humidity Appendix C - Differential Pressures Appendix D - Finite Element Formulation Appendix E - Computational Fluid Dynamics (CFD) Analysis Appendix F - Selected Deformation Results Appendix G - Wind Speed, Direction and Pressure Appendix H - Acceleration/Vibration Results Appendix I - Stress Analysis for Various Lifting Scenarios


List of Tables

Table 1: Temperature patterns throughout the experiment........................................................... 25 Table 2: Heat transfer properties................................................................................................... 36 Table 3: Constants for sorption isotherm function (ASHRAE, 1997; Richard, et al, 1997) ........ 37 Table 4: Peak deformations at various process stages (mm) ........................................................ 55 Table 5: PPV and RMS data ......................................................................................................... 59 Table 6: Vibration induced damage level (from Splittgerber 1978)............................................. 59 Table 7: Properties used for the analysis ..................................................................................... 64 Table 8: Comparison of deformations (mm) recorded during field tests and model predictions . 66 Table 9: Comparison of deformations (mm) recorded during field tests and model predictions 66 Table 10: OSB Strength Properties (Mi, 2005) ............................................................................ 69 Table 11: Stress factor k values for elements 15479 and 8541.................................................... 69 Table 12: Scenarios for sensitivity study on lifting arrangements................................................ 72 Table 13: Comparison of stresses in plasterboard for lifting scenarios Cases 1 and 3................. 73


List of Figures

Figure 1: The test building............................................................................................................ 18 Figure 2: Building during construction and its panel-joint locations ........................................... 18 Figure 3: Wall layers..................................................................................................................... 19 Figure 4: Sensor location and data acquisition system ................................................................. 21 Figure 5: Hygrothermal sensors used in the project ..................................................................... 22 Figure 6: Wind sensors ................................................................................................................. 22 Figure 7: Indoor temperature ........................................................................................................ 24 Figure 8: Temperature at mid-height of the wall panel-to-wall panel joint (Channel 21)............ 24 Figure 9: Outdoor temperature...................................................................................................... 25 Figure 10: RH at the top of the wall ............................................................................................. 26 Figure 11: Outdoor Relative Humidity ......................................................................................... 27 Figure 12: Correspondent position N, S, E, W and degrees and building orientation.................. 27 Figure 13: Numbering system for the differential pressure sensors ............................................. 28 Figure 14: Differential pressures at the top of wall cavity (clear wall) ........................................ 29 Figure 15: 2-D domain wall assemblies........................................................................................ 34 Figure 16: Typical sorption isotherm curve for building materials (Richards et al, (1992))....... 37 Figure 17: Finite element and finite difference comparisons ....................................................... 39 Figure 18: Moisture content in storage layers............................................................................... 40 Figure 19: Predicted temperature and moisture distributions across the wall .............................. 40 Figure 20: Geometry and FE mesh of the wall assembly ............................................................. 41 Figure 21: Predicted temperature and relative humidity for one-year period of the sensors shown

in Figure 20........................................................................................................................... 42 Figure 22: Temperature and moisture histories near the wall panel-to-wall panel joint .............. 44 Figure 23: Types of wall panel-to-wall panel joints ..................................................................... 45 Figure 24: Comparison of moisture distribution between two panel joint systems...................... 46 Figure 25: Prefab mini home tested .............................................................................................. 51 Figure 26: Sensor placement (A&B=deformation, P= pressure taps) .......................................... 53 Figure 27: Computer data acquisition system............................................................................... 53 Figure 28: Damages observed....................................................................................................... 56 Figure 29: Deformation around a window corner (sensor A10) during factory lifting ................ 57 Figure 30: 3-D finite element model of a prefab mini home ........................................................ 60 Figure 31: Lifting position ............................................................................................................ 60 Figure 32: Modelling boundary supports during lifting ............................................................... 61 Figure 33: Gaps between floor beams and flat bed....................................................................... 62 Figure 34 The arrangement of mini home on the flat bed of the tractor-trailer............................ 62 Figure 35: Modelled support conditions during transportation .................................................... 63 Figure 36: Connection details modelled using link elements ....................................................... 64 Figure 37: Deformed shape of mini home during lifting.............................................................. 66 Figure 38: Effective stresses (Von Mises stress contours, eσ ) at location of sensor A10 (MPa). 67 Figure 39: Von Mises stresses near corner of door opening......................................................... 70 Figure 40: Double rim and single rim headers for floor framing ................................................. 71


1. Goal and project objectives

The goal is to ensure that the Canadian wood industry possesses appropriate technologies to adequately compete in a growing sector of the construction industry. The specific objectives are:

• To help maintain wood products as a viable construction material within a dynamic building industry and regulatory setting.

• To promote a factory based manufacturing approach to building with wood. • To promote efficient and appropriate use of wood-based construction materials. • To increase the proportion of construction that is modular rather than stick-built. • To improve quality control with respect to performance objectives for completed structures. 2. Introduction Prefabricated (also known as prefab or modular) wood construction is an important part of the Canadian building industry. Despite that the prefab systems currently comprise only about 10% of new light-frame houses constructed in North America, because stick-built systems still dominate, it is clear that the industry will play an increasingly important role in coming years. A recent study sponsored by the Canadian Mortgage Housing Corporation reports that consumer acceptance of prefab systems for housing, and other purposes, is rising and that production will continue to turn towards ‘high end’ designs (CMHC, 2005). Prefab construction manufacturers in North America produce complete buildings, or modules or panels for buildings, in a controlled factory environment rather than stick by stick on site. Prefab homes and other buildings can be fabricated with consistent quality compared to stick-building. Quality control is maintained by in-house inspection throughout the construction processes. Other inspections by a third-party agency are usually conducted both at the manufacturing plant and the building assembly site to assure that production, standards, methods, and materials meet required standards. Cost is also a driving factor for modular/panellized construction to compete with the stick-built construction, since the manufactured housing companies can bulk purchase construction materials, achieve assembly line labour efficiencies and minimise material wastage. Prefab also promotes safer working practices. In general there are four types of factory-built construction, listed in order of their completeness of prefabrication: 1) manufactured systems (formerly known as mobile or mini homes in Canada and USA), 2) modular systems, 3) panellized systems, and 4) pre-engineered/pre-cut systems. The following are descriptions of each of these systems. Manufactured (mobile/mini) systems: Manufactured homes are the most complete prefabricated products. Homes or other small buildings are typically complete with interior and exterior finishes, and plumbing, electrical and mechanical systems. In addition, they are typically built with an integrated frame that allows them to be transported to site using axles or bogeys and placed onto pre-constructed foundations. Currently factory production lines are semi-automated


with a usual fabrication sequence being: 1) construction of a floor platform, 2) installation of wall panels, 3) installation of roof or ceiling, 4) addition of exterior elements, and 5) interior finishing. In general for housing buildings employ ‘heavier’ construction techniques than equivalent stick-built construction, so that they can resist forces during handling, transportation and site erection/installation processes. The most commonly applicable building codes are CSA Z-240 for manufacturing processes and The National Building Code (of Canada): Part 9, that apply to residential housing design and construction. City or Provincial codes override these regulations in some jurisdictions. In the USA the most common building code used is the Federal HUD Code for Manufactured Homes that can supersede local building regulations. Modular systems: Like manufactured homes, modular homes/small buildings may have a steel undercarriage during transportation that is generally not a permanent or necessary structural component. Modules have been used to create buildings up to several storeys high, but that is unusual. Most buildings are one-storey houses consisting of 2 to 3 modules or two-storey houses consist of 4 to 5 modules. Factory production lines are similar to those for manufactured homes. In Canada there is a developing trend towards constructing multi-residential housings up to four storeys high. Panellized systems: Panellized systems more labour intensive on the construction site than manufactured or modular systems because there is no preassembly creating a three dimensional arrangement. However, they provide a degree of design flexibility even in standard sizes. In panellized systems, panels are usually produced in sizes that suit certain building styles and configurations. The panels themselves fall into classifications of ‘open’ or ‘closed’. Open panels refers to factory-assembled wall, floor or roof panels that are open on one or both sides to facilitate construction, and installation and regulatory inspection of mechanical, electrical and plumbing services. An exterior open panel wall may have sheathing, doors, windows, and siding on the outside and insulation between the studs, but will lack finished materials such as drywall on the inside surface. Closed panels are enclosed on both sides. Panel factories are often part of manufactured or modular homes companies. Due to efficient packaging practices, panellised products are easily by road or rail or exported overseas by ship. Pre-engineered/Pre-cut systems: Pre-engineered or pre-cut systems are the least prefabricated and most expensive type of factory-manufactured buildings, and require the most on-site work. The name is something of a misnomer because systems consist of precision pre-cut components that will fit together like jigsaw pieces to create a so-called pre-cut system. They come in a variety of types including post-and-beam construction, log-homes, a-frame buildings and geodesic domes. Normally pre-cut components are bundled in system lots for sale. They are the simplest type to ship and often there is little wasted material. Despite some advantages over stick-built construction, prefab construction faces some technical and non-technical barriers. Prefab manufacturing in North America tends to be very fragmented with individual companies focussed on manufacturer of their proprietary designs (NAHB, 2002). This approach may appear beneficial from a local standpoint, but it is not helpful to rapid expansion of the industry or cost effective automation of plants. Also, the typically small corporations do not have resources to support R&D necessary to improve their products and to move them to higher or new market niches. Other factors that can impede Canadian prefab


builders include different applicable code requirements in alternative markets. For a specific market like New England states of the USA, the Canadian prefab products do not have significant problems meeting local codes. However, for other regions, such as southern parts of the USA, meeting regulatory requirements against potentially damaging effects of wind and rain, and high external ambient humidity can be daunting. A key issue for the industry is being able to handle different performance for different markets. Prefab systems made in Canada must be able to be customized in order to expend their marketability outside of local or traditional markets. Transportation/shipping by road is a major issue for modular/mini home companies in some regions of Canada and USA because of strict regulations on the maximum size of home-modules that tractor-trailers can carry. Typically modules 13½ feet (4.11 m) or 16 feet (4.88 m) wide may be transported. Special safety measures have to be taken during transporting. Modules must be designed to resistance high dynamic stresses arising during trucking and site-installation. To accommodate this, some modular manufacturers use up to 30% more lumber than is required by prescriptive regulations applicable to stick-built structures (PATH Inventory, 2003). This can negatively impact the efficiency and affordability of prefab systems. From a structural point of view, making the framing system too stiff or wrongly placing reinforcing materials is bad practice, because it can cause modules/components to become more rather than less susceptible to damages during handling and transportation. Often damages are usually minor and do not impact the structural performance significantly. However, damage is often highly visible as cracks in walls and ceilings. Presence of any damage to new units leads to perceptions that prefab buildings are low quality products. Also, damage effects long-term performance in terms of factors like air leakage characteristic and durability of materials that are the fabric of the buildings. Special design methods are needed to eliminate possibilities that units will be damaged prior to service and that once in service they will have high performance and will not deteriorate prematurely. Despite panellized systems being in the greatest demand, among alternative types of prefab systems, there are important performance problems associated with the construction joints that exist where panels are jointed (Traynor, 2007) integrity. It is the nature of panellized system to have construction joints between components (panels) and because of the modularization involved they are normally less discrete than in stick-built construction. Without careful design and construction air can leak through those joints leading to wasted energy, uncontrolled moisture depositions, and deterioration of wood and other material that comprise the panels. It is therefore important to design and construct prefab panels that meet acceptable structural, energy, and durability performance levels. Based on the above background, and discussions with representatives of the prefab industry and building regulators in Canada, it was decided that the project reported on here would focus on two important aspects of prefab construction. The first sub-project was to investigate the building physics aspect of performance, with emphasis on air leakage and moisture deposition in prefab panellized wood buildings. The second sub-project concerns structural engineering aspect with emphasis on performance of prefab mini/modular wood buildings during handling and road transportation processes.


3. Overview of method

The research was a combination of field and laboratory experiments, and numerical modelling. Data collected during experiments was used to develop and verify the numerical models. Study of technical and scientific literature helped guide selection of field experiments and provided inputs to the numerical models. Major experimental activities were full-scale in-situ tests of actual prefab buildings. Advanced computer data acquisition and manipulation systems facilitate collection and handling of immense data sets from experiments. The following are brief descriptions of the research method, with full details contained in subsequent sections and appendices of this report: Part I - Building Physics Aspect: The end goal is to identify ways of improving control of air leakage and moisture performance of construction joints (interface details) employed in panellized light-frame wood buildings. Research activities covered long-term field monitoring of a UNB owned building with panellized walls and development of numerical models of the physical phenomena. The long-term field monitoring tracked moisture accumulation in wall components and correlated it to air pressure (leakage) flow within the wall at locations with and without a vertical construction joint. Observations included the effects of seasonal weather variations. Several hygrothermal (moisture) sensors were installed at locations of interest. These included pressure differential sensors, temperature sensors (thermocouples), relative humidity sensors, and an anemometer for measuring the ambient wind speed and direction near the building. These sensors were connected to a data acquisition system programmed for a long-term capture, storage and downloading of data. Monitoring was carried out between spring of 2005 and winter of 2007. Data from field measurements and component performance data collected by other researchers (e.g. NRC-IRC), were used as the basis for developing numerical models. Models employed so-called finite element methods to quantitatively describe heat, air and moisture transfer in wall assemblies, including joints in panellized walls. The developed models were verified using the field-monitoring data. Once the models were verified to be robust, several alternative construction joint methods were investigated numerically with respect to their air leakage and moisture performance. This yielded recommendations for the manufacture of walls with better performance than is currently typical. Note: It was intended originally in this project to conduct a long-term field measurement of an occupied prefab wood building in collaboration with a homeowner and a prefab manufacturer. However, this proved unnecessary because data collected for the UNB-building was sufficient for the model verification. Part II - Handling and Transportation Aspect: The purpose of the work was to identify procedures and techniques that minimize the possibility of mechanical damage to modular home sections or complete manufactured/mini homes during their factory or site handling or during their transportation from the manufacturing plant to a construction site (or retail location). The concept is that if the units themselves and the handling and transportation practices are appropriately engineered, instead of being based on trial and error, it will be possible to both minimize potential for damage and reduce the amount of construction materials.


Activities encompassed field experiments and numerical modelling. In coordination with a local modular housing company in New Brunswick, the project team developed a special system for measuring forces and deformations on prefab units/modules during lifting and transportation processes. Four important structural engineering parameters measured during transportation were wind speed and direction with respect to the truck axis; wind forces on surfaces of modules; deformations in modules; and forces due to road roughness, cornering and acceleration and deceleration of the tractor-trailer. During lifting/handling processes at the manufacturing plant static deformations around wall openings were measured because those openings are commonly loci for damage. Damages generated were examined and correlated with observed deformations. Two computer data acquisition systems (DAS) were used. The first DAS captured low frequency signals from an anemometer (measuring wind speed and direction), pressure taps (pressures on module surfaces), and deformations in the module. The second DAS captured high frequency signals of accelerations at various locations on the module and truck. Data collected was analyzed to understand structural behaviour of prefab buildings during handling and transportation. Finite element analysis was used to create structural models for the performance of prefab modules during lifting and transportation. The 3-D models incorporated all-important components of the construction. Orthotropic shell elements were used to model wall and roof sheathing, drywall (plasterboard), and floor sheathing. Wall studs and roof truss members were modelled using orthotropic shell elements, instead of common frame elements, because that accommodated investigation of force distributions between sheathing and plasterboard. Floor joists and roof framing were modelled using common frame elements. The interfaces / connections between framing members and framing and other components were modelled using link element composing of internal springs (axial, shear, and rotational). Major loads imposed on building modules included dead loads, wind excited forces, and ‘road vibration’ forces and other acceleration generated forces measured during field tests. Structural stresses at critical building features like window/door openings and framing junctions / connections in walls were analysed and failure criteria applied to assess the likelihood of damage. Once verified models were used to define and demonstrate potentially effective damage mitigation strategies concerning structural detailing and how modules should be supported during lifting and transportation operations.


PART I

Building physics performance aspect of panellized wood buildings


4. Building physics performance aspect of panellized wood buildings 4.1 Background Air leakage (or leakage for short), along with the choice of heating and ventilation/climate conditioning systems, are key factors in energy use and conservation, and creation of a health indoor climate, irrespective of the form of construction. Leakage in buildings is caused by intentional openings and permeability or unintentional damage such as cracks in the building enclosure/envelope. In prefab wood system leakage usually occurs at the construction joints/interfaces between sub-assemblies. Such joints that are typical for prefab light-frame wood construction are not the same as those in stick-built light-frame wood construction. Apart from leakage and waste of heating energy, poor joints can cause excessive moisture deposition in building components. Avoidance of moisture deposition within the fabric of buildings is important from several points, with the most important being:

• Excessive moisture creates host conditions for biotic degradation of structural and non-structural wood-based and other materials,

• Excessive moisture creates host conditions for fungal growth/moulds injurious to human health.

• Failure to control moisture content or moisture movement can lead to impaired serviceability of buildings due to deformations in components and subsystems.

Wood and wood-based engineered composite products are the primary components in prefab building assemblies in North America and other regions. Wood as a hygroscopic material shrinks and swells as moisture content (MC) in it changes. MC is a measure of how much water is in any piece of wood relative to the wood substance (i.e. weight of water in the wood divided by the weight of that wood in an oven dry condition). As any hygroscopic material, wood placed in an environment with stable temperature and relative humidity (RH) will eventually reach a moisture content that yields no difference in vapour between the wood and the surrounding air. MC will therefore stabilize at a point called the equilibrium moisture content (EMC). Wood used indoors will stabilize at an EMC of between 8 and 14 percent, and outdoors at an EMC between 12 to 18 percent if there is not direct wetting of it. Wood’s hygroscopic nature is not necessarily a bad thing, and during a building’s service life wood parts function as a natural humidity control device. Wood is either a sink or reservoir for moisture from a building’s atmosphere, depending on the sign of any differential in vapour pressure at its surface. Usually, when the indoor air is very dry, wood will release moisture, and when the indoor air is too humid, wood will absorb moisture. To reduce air leakage and moisture related issues, it is crucial to understand the long-term hygrothermal behaviour of light-frame wood building envelopes. This is done through combinations of field or laboratory experiments, and computer simulations. Robustly verified computer models can be an invaluable tool for predicting behaviour of building configurations that have not been studied experimentally. Gaining understanding through experiments alone is usually not straightforward and is time consuming and very costly. The following sub-sections present field-monitoring using a panellized wood building belonging to UNB, and the development of numerical models of heat and moisture transfer in building assemblies.


4.2 Long-term field monitoring program The main purpose of the field monitoring was to investigate air leakage and moisture performance of the panellized building over the course of seasonal weather patterns. The derived data therefore yielded information about a wall’s response under a very wide range of combinations of indoor and outdoor climates that both influence drying or wetting rates, as well as a realistic insight into diurnal and seasonal effects. Correct choices of types and locations of monitoring sensors, and the method of interpreting data, was critical for successful field experiments. The collection of data such as wind speed and direction, differential pressure across wall layers and interfaces, temperatures and relative humidity provided information to assess heat loss and moisture movement, and moisture deposition in the building enclosures and particularly insulation filled cavities in wall assemblies. Although similar data has been collected by others for stick-built walls, none had been collected in a manner that recognizes the unique features of panellized wood buildings. 4.2.1 Related field-monitoring activities The phenomenon of moisture deposition in any building envelope is complex and involves interaction among moisture (hygrothermal) parameters such as temperature, RH, differential pressures between indoors and outdoors, and solar radiation. Historically, research on moisture related issues had focused on controlling diffusion processes from the indoor to outdoor by providing vapour retarders (vapour barriers) and limiting water entry from the outside by providing siding to deflect rain water. However, recently air movement (leakage) has become an important consideration, especially as energy costs and concerns about conservation of resources and production of greenhouse gases rise. Air leakage and moisture problems in North American buildings was documented as early as the1960`s, and since then there have been many research studies and field monitoring activities. Most early field measurements addressed issues related to comparison of performances of various wall assemblies exposed to actual weather patterns. For example, field experiments by Johnson (1982) addressed whether wall assemblies situated in north-east and north-central USA could perform well without vapour barriers being installed. In his study, 17 wood-frame homes were monitored during the month of March 1977, which was a month when high MC accumulations were expected. All homes monitored except one had vapour barriers in the walls. Results indicated that all wall frame assemblies with vapour barriers had better moisture performance than the wall assembled without a vapour barrier. The measurements made were straightforward and no large scale and long term monitoring involved. A more elaborate field monitoring exercise was conducted by Tenwolde et al (1995). In his study field measurements on an experimental wood building (stick-built system) were performed using a programmed computer data acquisition system. The main data collected was pressure differential in the wall envelope during the heating season to provide information about the quantity of air leakage and direction of flow. There was also a few relative RH data collectors installed in the wall to examine the condensation process. The huge amount of differential pressure data collected made air leakage analysis difficult, and the analysis was only performed for certain short time


periods. In general the results suggested that airflow direction in a building is influenced by the wind speed and direction or air around the building. The study also concluded that excessive leakage could lead to moisture related problems (MC high enough to support mould or fungi) and that the indoor relative humidity was a key factor for control and prevention of excessive moisture accumulation in building envelopes. More recent field monitoring studies of light-frame buildings have been reported by Straube et al (2002), Porter (2003), and Rose and Francisco (2004). It can be concluded from all past field studies to achieve reliable results it is essential to carry out hygrothermal measurement over a long period, which requires reliable sensors and good data acquisition systems. Measurements must be accurate and free from error in temperature, airflow observations and moisture that affect each other. However, despite knowing this, so far there has been no broad consensus on field test standards and how to interpret complex data that is collected. The issue becomes more complex if new construction methods and materials are used in buildings, which is now a common circumstance. There seems to have been a gap between accuracy of analytical and laboratory studies, and practical field observations. The research team for work reported here made great efforts to avoid that pitfall and what is described below reflects this. Detailed knowledge of prefab building wood buildings is also required because that effects instrumentation decisions. 4.2.2 Test building The test building is located in the compound of the UNB Tweeddale Centre at the Hugh John Fleming Forestry Complex in Fredericton, New Brunswick. This site has an elevation of 370 ft. (110m) above sea level, and the building latitude is 45°55’ N and longitude 66°39’W. According to Environment Canada, its ID climate is 81016000, which is part of the Atlantic climate zone. The building is oriented such that the prevailing wind direction is approximately perpendicular to the longitudinal axis of the building. The building has a construction typical of bungalows in Canada and the USA designed using the so-called 2x4 system (Figure 1). It is a platform-type framework constructed using panellized wall segments with a length of approximately 3.6 m. The building dimension is 26’x 51’ (8.5m x 17m). During the hygrothermal experiments, the interior of the building was only partly complete with no wall partitions installed and only partially lined with plasterboard. There was non-mechanical aeration of the indoor space. The building had only two openings in the form of pedestrian doors and no trapdoor ventilation. The floor platform sits on a frost wall with a concrete foundation 10” (254mm) thick and having an outside wall height 16” (406mm) above grade. A crawlspace is beneath the floor is 4’ (1.3 m) high and has no screened ventilation, i.e. air inside it is not in movement. The floor consists of 42 free-span wood I-joists spaced at 16” (0.406m) on centre. The floor sheathing is composed of 4 x 8 ft (1.22 x 2.44m) OSB sheets with 5/8” (15.9mm) thickness. No floor finishing had been installed. The roof consists of 29 pre-manufactured free-span gable roof trusses spaced at 2’ (0.61m) on centre. The attic has ventilation baffles at the eaves near each truss, along the long side walls. During the field monitoring the building was unoccupied, with temperature and RH controlled manually using thermostats to normal domestic standards rather than laboratory precision. In


addition to this hygrothermal monitoring, the building is also instrumented with many load cells to examine flow of forces (load-paths) in the building components due to environmental loads like wind and snow. The work on load path project is beyond the scope of this NRCan project but is reported on elsewhere (Smith et al 2006).

Figure 1: The test building

Hygrothermalsensor locations

: panel-to-panel (butt)joints

Hygrothermalsensor locations

: panel-to-panel (butt)joints

Figure 2: Building during construction and its panel-joint locations


4.2.3 Test wall All walls of the building have nominal 2x4 (38 mm x 89 mm finished) Spruce-Pine-Fir (S-P-F) framing for studs, and sole and header plates, with nailed on OSB sheathing. The studs are spaced at 2 ft (0.61 m) on centre. Three of the walls were assembled based using open prefab panellized wall segments (wall panels) and employing a 12 ft (3.66m) module. The fourth wall was stick built. Construction details are simple throughout, including vertical joints between adjacent wall panels. For walls that are panellized panels are butt-jointed and fastened together with interconnected with nails. No caulking or gasket material is inserted in those joints. These practices are typical of construction practices in Canada. Following erection of the walls fibreglass insulation was installed in cavities between studs, air and vapour barriers installed, plasterboard lining installed on a selective basis, and external rigid insulation and drop siding applied externally everywhere. Figure 3 summarizes the construction. Materials within the walls of the test building are:

• Framing: planed stud-grade S-P-F, • Sheathing: 4 x 8 ft (1.22 x 2.44m) OSB panels 7/16 in (11.1mm) thick, • Interior lining: 4 x 8 ft (1.22 x 2.44m) drywall sheets 0.5 in (12.7mm) thick, • Sheathing-to-framing connection: 2.25 in (57.2mm) long common nails spaced at 6 in

(152.4mm) along panel edge connections and 12 in (304.8mm) along interior connections. • Stud-to-plate connection: 3.5 in (88.9mm) spiral nails (2 per joint). • Plasterboard-to-framing connection: drywall screws spaced at 300 on centre. • Integral insulation: fibreglass matt in framing compartments. • Exterior insulation: Styrofoam panels 24in x 96in (600mmx2400mm) attached to outside the

OSB sheathing using nails spaced 300 on centre. • Siding: drop siding with stained wood nailed on vertical wood strapping spaced at 2 ft

(0.61m) of the centre.

Figure 3: Wall layers

Fibe

r gl

ass i

nsul

atio

n

Rig

id fo

am in

sula

tion

OSB

Gyp

sum

boa

rd

Vap

our

barr

ier,

2 m

il

Bui

ldin

g p

aper

Air

/spa

ce (f

or st

rapi

ng)

Woo

d si

ding

38 89

25 11 12.7

15

Fibe

r gl

ass i

nsul

atio

n

Rig

id fo

am in

sula

tion

OSB

Gyp

sum

boa

rd

Vap

our

barr

ier,

2 m

il

Bui

ldin

g p

aper

Air

/spa

ce (f

or st

rapi

ng)

Woo

d si

ding

38 89

25 11 12.7

15

Note: Gypsum wall board (GWB) and plasterboard are used interchangeably in this report


4.2.4 Field sensors and other instrumentation The hygrothermal parameters measured in this project included temperature, relative humidity, differential pressure, and exterior wind velocity and direction. The instrumentation was designed to enable comparison of the hygrothermal behaviours between vertical wall construction joints (butt joints) between adjacent wall panels, and the hygrothermal behaviour of clear walls. The chosen locations were for a segment of wall that had an interior plasterboard lining. In discussion that follows the term ‘joint’ implies a construction butt joint. The terminology clear wall means the location of a wall compartment (wall cavity) in the centre of a wall panel. Sensors were positioned in two different locations. One set of sensors was positioned at the middle of a wall panel and another set was located at a joint. The ‘joint between plasterboard sheets’ was offset from the construction joint between wall panels. The test arrangement is shown in Figure 4. Six thermocouples were installed just behind the OSB sheathing to monitor the potential dew point temperature. The dew-point temperature is the temperature at which moisture will begin to form on a building surface. It is the temperature to which a volume of air must be cooled in order to reach saturation. It is common known that in cold regions (heated indoor building climates) that OSB sheathing is a vulnerable place with respect to condensation that ultimately causes moisture related problems. Four relative humidity sensors were installed near the OSB sheathing and were distributed at the top and middle part of both the cavity of the clear wall and at the joint. In total there were 18 differential pressure sensors installed. Three differential pressure sensors were installed at each of the top, middle and bottom of the wall for both the cavity and joint. For each set of sensors the first was positioned just behind the plasterboard, the second was behind the OSB sheathing and third was just behind rigid foam insulation. In this way, the airflow (air leakage) can be detected by examining the relative differential pressure recorded in each position. All sensors were linked with a computer data acquisition system that recorded their measurements at either one or five minute intervals. There were no direct moisture measurements made. MC was estimated based on RH and temperature records and using well known sorption isotherm relationship curves. Detail specifications of each sensor are described below. Temperature measurements: The device for temperature measurements is a thermocouple (Figure 5a). Thermocouples used in this project were of the J-type which has a positive iron wire and a negative constantan wire. The temperature fluctuation in the measured region/surface is determined based on the thermal expansion change of the wires. J-type thermocouples are suitable for measurements in the range between -40ºC and +200ºC with a resolution of ± 0.5ºC. Relative Humidity measurements: To measure RH omega-type sensors were used (Figure 5b). These devices can make measurements between 0% and 100%RH, with an accuracy of ±2.5%RH between 10% and 90%RH and ±5%RH for <10% and > 90%RH. Readings are sensitive to temperature variations below -5ºC and above +40ºC. During the monitoring activity, the lowest temperature recorded


was about –-5ºC and the highest temperature was +30ºC. Therefore, the sensor type was appropriate for the monitoring carried out. Before the sensors were used, they were calibrated using chemical (salt) solutions that produce different air RH at various temperatures.

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

10

11

12

10

11

12

16

17

18

16

17

18

1

2

3

1

2

3

4

5

6

4

5

6

7

8

9

7

8

9

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m

Figure 4: Sensor location and data acquisition system

AA

Cross-section A: Thermocouples: Relative humidities: Differential pressures

: Differential pressure tubes,connected to pressure sensors

: Thermocouples: Relative humidities: Differential pressures

: Thermocouples: Relative humidities: Differential pressures




Pressure sensors: Measurements of the pressure differences are useful for assessing the air leakage potential of building constructions, pressure moderation performance, and ventilation potential. Low pressures variation in the range 0.5 to 5 Pa can move significant amounts of air. In this project the 160PC series sensors made by Honeywell suitable for low-pressure type measurements were used (Figure 5c). These sensors have two ports that allow connection to two pressure sources. The sensor has a pressure range ±5.0 in of H2O (1250 Pascal). According to the specification, the sensor’s operating temperature range is between -40°C and +85°C, which was appropriate for the project.

(a) Thermocouple (b) RH (c) differential pressure

Figure 5: Hygrothermal sensors used in the project Wind speed and direction: Pressures in building envelopes vary with exterior wind speed, and also between locations where the wind hits the building envelope indicating that knowing the wind direction is also important. The wind sensors used were two anemometers of type 5103 manufactured by RM Young (Figure 6a). Each wind speed sensor is a four-blade helicoids propeller with its rotation producing an AC sine wave voltage signal that can be proportionally related to wind speed. Vane angle is sensed by a precision potentiometer housed in a sealed chamber. With a known excitation voltage applied to the potentiometer, the output can be proportionally related to vane angle. The wind speed range that can be measured is between 0 to 60 m/s (134 mph) with gust survival 100 m/s (220 mph). The wind speed accuracy is about ±3 m/s (±0.6 mph) and the wind direction accuracy ±3 degrees. In this project two anemometers were mounted near the building at 5m and 10m heights above ground level (Figure 6b). The recorded values were the average between the two recording heights.

(a) anemometer (b) anemometer location

Figure 6: Wind sensors

10 m

5 m


4.2.5 Monitoring and recording Ideally, the hygrothermal parameters need to be measured at frequencies that depend on the natural responses time of the variables involved. For example, temperature and relative humidity are ideally measured at one-hour interval because their values normally change on an hourly time scale. Differential pressures are ideally measured every second, because they are affected by fluctuating wind gusts at building surfaces that alter in seconds. However, to reduce the amount of data recorded, in this project the monitoring frequency for all sensors was the same. At first readings were monitored every five minutes but then due to occurrence of strong winds the rate was changed to every minute. Afterwards it was changed back to every five minutes. The computer data acquisition program used Strain Smart from Inter Technology. The data recorded was transferred to another computer for processing. The monitoring activity was started on April 2005 and ended Feb 2007. During early monitoring period wood siding was not installed. As was anticipated that affected, to some extent, the pattern of record data for hygrothermal parameters. 4.2.6 Field test results All results were plotted using the computational software MATLAB. Results included temperature, RH, differential pressures, wind speed and direction as averaged hourly values in a 22 month period. Plotted parameters are shown in Appendix A (temperatures), Appendix B (relative humidity), and Appendix C (differential pressures, and wind speed and direction adjacent to the building). Each graph in those appendices was divided into two time periods. The first set represented one-year from April 2005 to April 2006, and the second a ten month period from April 2006 to February 2007. The first set of graphs was used as verification of numerical hygrothermal models discussed later in this report. 4.2.6.1 Temperature As shown in Figure 4, the sensors measuring indoor room temperature were Channels 27 and 28. The sensors located in the construction joint were Channels 24, 21, 22 which had respectively top, middle and bottom locations. Sensors in the wall cavity were Channels 25, 23 and 26 at respectively top, middle and bottom locations. Figures 7 and 8 illustrate recorded temperatures for the indoor and middle-height wall joint locations during the one-year period (Apr 2005-Apr 2006). During the spring and summer times (indicated by the months numbered 1 to 5 on the horizontal axis) when the thermostat temperature was set at off, the indoor temperatures varied considerably from the high daytime temperature of about 25°C to the lower night time temperature of about 12°C (Figure 7). This reflects that there was no air conditioning system in the building. Under winter conditions as the temperature outdoors dropped down the indoor temperature had a steady average of about 17°C with small variations between daytime and night time. The same temperature pattern and range were observed in another indoor measuring temperature, Channel 28 (Appendix A).


Figure 7: Indoor temperature

Figure 8: Temperature at mid-height of the wall panel-to-wall panel joint (Channel 21)

May Jul Sept Nov Jan Mar



Figure 9: Outdoor temperature The recorded temperature pattern in the middle wall-to-wall joint shows a different pattern than the indoor temperature. Comparing Figures 8 and 9, the temperature at the joint follows that of the outdoor temperature fluctuation, with maximum temperature range (15 to 27ºC) during summer and a minimum temperature range (5 to 15ºC) during winter. As can be seen in Appendix A, despite different values in the temperature ranges, the same temperature patterns were observed in other channels including 24, 22 (joints) and 25, 23 and 26 (wall cavity). The temperature variation was slightly higher at the tops of walls than in the middle and bottom locations. Table 1 summarizes temperature variations in the wall cavity, and wall panel-to-wall panel joint as seasonal patterns. Numerical values are assigned to these sensors, with value 1 indicating the location with highest average temperature and 6 that with lowest recorded values.

Table 1: Temperature patterns throughout the experiment

JOINT WALL Channel 24 Channel 25

Spring Summer Autumn Winter Spring Summer Autumn Winter TOP 3 2 5 5 3 1 6 6

Channel 21 Channel 23 Spring Summer Autumn Winter Spring Summer Autumn Winter MIDLLE

4 4 1 1 4 3 4 4

Channel 22 Channel 26 Spring Summer Autumn Winter Spring Summer Autumn Winter BOTTOM

4 6 3 3 4 5 2 2 The wall cavity sensors recorded higher temperatures than the sensors at the same height in the joint. Top locations were hottest in the daytime and bottom locations the coldest. In the joint the hottest point in was the middle in the autumn and winter, while in the wall cavity the hottest spot


was at the bottom. Temperatures recorded in the top positions (both at the joint and wall cavity) was the hottest during summer season, and the coolest during the winter. 4.2.6.2 Relative humidity The RH sensors in the wall cavity were labelled as Channels 36 and 39, and the sensors in the joint are labelled as Channels 19 and 20. Complete RH results can be seen in Appendix B. In general it can be noted that there is a strong relationship between the temperatures and RH in the wall assembly. During the summer season of the first monitoring period, RH both in the wall cavity and joint showed the maximum range of values (50 to 85%) whereas winter showed the minimum values (20 to 40%). There was a decrease of RH values for all locations during the second monitoring period as the wall dried out after initial construction, but the trend values are the same, during the summer (50 to 70%) and during winter (20 to 40%). Comparing Figures 10 and 11, it can be seen that high outdoor RH fluctuation did not significantly influence the RH pattern recorded within the wall assembly. In this project, indoor RH was controlled at around 50% throughout. The indoor RH control was routinely checked using direct RH measurement, and the actual values fluctuated between 45% and 55%. Among all RH sensors the top wall cavity sensor (Channel 36) recorded the highest RH values, which were between 50 and 80% during the first monitoring period and between 35 and 70% during the second period (Appendix B). When the temperatures cooled down, the RH sensors located at the top of the joint increase its difference with the rest of the sensors with the difference being about +10%. During the fall to early winter period the sensors with higher RH values were the sensors at the top of wall cavity and the middle of the joint. During the coldest period (winter) the sensor in the middle of the wall cavity (Channel 39) recorded the lowest RH values, while at the same time the sensor at the top of the wall cavity maintained values around 60%. The RH at the top of the cavity was about +10% relative to the other three locations.

Figure 10: RH at the top of the wall



Figure 11: Outdoor Relative Humidity 4.2.6.3 Wind speed and direction The wind speed range recorded during the whole monitoring period was between 0 and 60 km/hr, with the majority of wind speeds recorded between 5 and 10 km/hr. High wind was recorded mostly during the fall to winter period. The most dominant wind directions at the test building facility were coming from East and North East. Plots of wind speed and direction are given in Appendix C. The directions in degrees shown are shown in Figure 12.

(a) anemometer direction and degrees (b) building orientation

Figure 12: Correspondent position N, S, E, W and degrees and building orientation

WEST 2700

EAST900

NORTH 0-3600

SOUTH 1800

North

Sensors location

250

North

Sensors location

250

North

Sensors location

250


4.2.6.3 Differential pressures Figure 13 shows the numbering system for the differential pressure sensors. Sensors located behind the plasterboard layer have the lowest labelled numbers; sensors located behind the OSB sheathing have the middle label numbers; and sensors located outside the OSB have the highest label numbers and were installed before rigid foam insulation was added to the outside of the wall. For example, Figure 13 illustrates the sensors placed at the top of wall cavity with the channels numbered 10, 11, and 12, representing lower, middle, and top locations respectively. Appendix C shows detailed plots of all differential pressures in each location, together with wind speed and direction near the building. No wind speed and direction are plotted for the second monitoring period, because the wind sensor was removed.

Figure 13: Numbering system for the differential pressure sensors

As shown in Appendix C, the sensors were very sensitive to small pressure changes generated within the wall assembly. The differential lowest pressure value recorded was –5 Pa and the highest +20 Pa, but the majority of values were positive. Positive pressure differential means that pressure inside a wall cavity is larger than in the indoors. Changes in the differential pressure trends, or values in each location, were used to indicate if there was airflow (leakage) across the wall assembly. Airflow from indoor to outdoor is called exfiltration and from outdoor to indoor is called infiltration. Figure 14 shows an example plot for the differential pressure at the top of wall cavity recorded during the first monitoring period. The following notes describe patterns of pressure differential changes in each location. Wall cavity: TOP (Channels 10,11, and 12): Channel 12 recorded an almost constant differential pressure of 7 Pa during the whole monitoring periods, while Channel 11 showed high differential pressure fluctuations during early portions of the monitoring period (spring 2005) and after summer 2005 constant pressure differential in the range of 7 Pa (Figure 14). Channel 10 readings fluctuated during spring 2005 from low to high. In summer 2005 this channel had lower pressure values than the other two and started to fluctuate during autumn 2005. During late autumn 2005

Wall thickness

Lower number for channel

Higher number for channel

Middle number for channel Outdoor Indoor


and early winter 2006 the readings jumped up to 12 Pa and then stayed almost constant until the end of the first monitoring period. The pressure difference trends between Channel 10 and the other two indicate that there was change in the airflow direction from infiltration during summer 2007 to exfiltration during late fall and early winter. The differential pressure showed the same trend for the second monitoring period (Appendix C). Also, it can be noted that there were differential pressure fluctuations at the top of wall cavity. MIDDLE (Channels 13, 14, and 15): These had approximately the same differential pressures throughout the whole monitoring period (about 7 Pa). The minor disturbances recorded are thought to be the result of wind gusts. BOTTOM (Channels 16, 17, and 18): Channel 17 was dysfunctional as indicated by ‘out of range’ differential pressure values. During the whole monitoring period Channel 18 recorded constant differential pressure values in the range of 7 Pa, while Channel 16 showed fluctuations. Despite the slightness of differential pressure changes between Channels 16 and 18, it can be said that there were small flows of air there.

Figure 14: Differential pressures at the top of wall cavity (clear wall) Wall joint: TOP (Channels 1,2, and 3): As can be seen in Appendix C, sensors at this location had the greatest differential pressure fluctuations during the whole monitoring period, between –5 Pa and +20 Pa. The middle sensor (Channel 2) recorded the lowest values, compared to Channels 1 and 3 during the whole monitoring period. Differential pressures were correlated to the wind speed and direction. Pressure differentials were highest when the wind velocity was highest. The difference was about +7 Pa if the wind came from the west and it was lower if it came from the north. As a result, the airflow was infiltration when the wind was from the west and exfiltration when it came from the north (see Figure 12 for the building orientation and sensors locations). MIDDLE (Channels 4, 5, and 6): In the joint all the sensors recorded almost the same pattern and differential pressure values, between 6-7 Pa. There was a weak effect of wind speed and



direction on the pressure values. Easterly winds increased the pressure slightly. Westerly winds varied the differential pressures downwards. No significant airflow was encountered in this location. Relative to that of the middle of the wall cavity this location had slightly more differential pressure fluctuations. BOTTOM (Channels 7, 8, and 9): Most fluctuations in differential pressures occurred at the bottom of the joint, Channel 9, in the range 0 to +12 Pa. During early autumn 2005 Channel 9 was about 4 Pa lower than the other readings, indicating there was exfiltration. After summer 2006 the readings of this channel became lower than the other, two indicating there was infiltration. Relative to the bottom of the wall cavity, this location had slightly more pressure fluctuations. 4.2.7 Summary of the field monitoring results In general, it can be concluded that air movements (leakage) occurred mostly near the wall panel-to-wall panel construction joint, rather than in the wall cavity. This was as expected because of the construction detailing there. Despite this the indoor temperature could be kept relatively constant, particularly during the winter season, indicating that in a general sense the building was not badly insulated despite the leakage. Such air movement at the wall joint accommodated drying mechanisms and there was a good balance in moisture deposition at that location. Recent observations (early spring 2007) and moisture measurement in the wall assembly indicate that moisture contents in wall studs are around 12 to 15% and there are no visible excessive moisture accumulations. The following are summaries of observations for temperature, RH and differential pressures. Temperature:

- Locations where the temperatures had more variation and the difference between summer and winter values were greatest were positions at the top (near the roof).

- Temperature variations in the wall cavity were higher than in the wall joint. - Wall temperatures were highest during summer and lowest during fall and winter.

Relative Humidity:

- Relative humidity increased as temperature in the wall increased. - Places where the movement of air was higher had lower RH, such as at the wall panel-to-

wall panel joint because aeration dries wood surfaces. - The top of the wall had deficient aeration resulting in the moisture content at that level

being the highest. This could indicate incipient humidity and moisture deposition problems there.

NOTE: Air leakage at walls is a poor and expensive mechanism for initially drying and maintaining dryness of wood materials in walls. Thus it should not be construed that these results constitute any endorsement that the observed ability of construction practices typical of current practice are ‘good practice’ in an overall sense.

Differential pressure and wind:

- Sensors located in the wall cavity indicated that at the bottom and at the top there was airflow from the inside to the outside (exfiltration).


- Differential pressures in the wall cavity at middle and bottom locations were slightly affected by the external wind direction and the wind speed.

- No observable effect of external wind on differential pressures was observed at the top of the wall cavity.

- Sensors in the wall panel-to-wall panel construction joint were more sensitive to wind speed and even more to the direction of the blowing wind at all the locations studied. The main direction of flow is exfiltration

- Airflow in the wall cavity was more dependent on temperature difference between indoor and outdoor climates, and the wall joint was more affected by the wind pattern than the wall cavity.

4.3 Numerical modelling of heat and moisture transfer Investigating hygrothermal performance of building assemblies through long-term field experiments is very expensive and time consuming. Also, field experiments are usually limited to a specific building system or configuration of components. Building designers need to have more general understanding and access to analytical/numerical tools that predict performance of alternative building configurations and construction details. Traditional analytical methods for predicting heat and moisture transfer are usually limited to one-dimensional problems with steady-state boundary conditions. Using such methods it is very difficult to investigate performance of construction details such as construction joints in prefab wood buildings. Therefore the goal must be creation of advanced analytical technique that utilize recent advances in computer hardware and software and implement realistic descriptions of physical systems and processes. Numerical modelling of heat and moisture transfer is not new in the fields of mechanical and chemical engineering. The basic principle lies in understanding coupled transport phenomena of heat (energy), air and mass (moisture) in porous materials. Despite much past study of heat and moisture transport in materials themselves, heat and moisture transport phenomena in building assemblies is far from being well understood, particularly with respect to multi-scale modelling issues. As physical processes heat and moisture transfer occur at the microscopic scale of porous building materials. However, the majority of numerical models assume the physical processes take place at the macroscopic scale, due to difficulty in measuring microscopic properties and extrapolating behaviours between scaling levels. Existing numerical models for building envelopes include hygIRC created by the National Research Council of Canada’s Institute for Research in Construction (NRC-IRC), MOIST created by the US Forest Products Laboratory (FPL), WUFI created by the Fraunhofer Institut of Bauphysik (IBP-Germany), and EMPTIED created by the Canadian Mortgage and Housing Corporation (CMHC). Most of these models replicate one-dimensional or two-dimensional domains and predict hourly hygrothermal parameters such as temperature and RH profiles across wall or roof assemblies as a function of pre-selected indoor and outdoor boundary conditions. Such calculations predict likely in-service performance over several years. For day-to-day design, one-dimensional models are usually regarded as being adequate. Simple models are mainly developed using finite difference or finite volume numerical schemes. The building


assembly is assumed to be simple, free from complex geometry or boundary conditions and materials homogeneous and continuous. Thus, realistic features such as electrical outlets in wall layers, construction joints in prefab wall panels and studs in light-frame walls are ignored. To handle complex geometries and boundary conditions that exist in real buildings in a realistic manner, it is necessary to modify current algorithms. In this project a two-dimensional finite element scheme was developed with the expectation that the resulting numerical models can achieve the objective of replicating unsteady (time-dependent) heat and hygrothermal (moisture) performance of construction joints in panellized walls of light-frame wood buildings. General assumptions in the governing equations and boundary conditions for heat and moisture transfer were made as close as possible to the physical phenomena involved. The models developed were verified using the field monitoring results already described in this report. Once verified the models were used to answer ‘what-if’ scenarios associated with prefab wall-panel-to-wall panel construction joints. In this respect, temperature, RH and moisture accumulation patterns are the model outputs of primary interest. 4.3.1 Governing equations Wood frame building envelopes are constantly exposed to both indoor and outdoor climate changes. The humidity difference between the building envelope and surrounding air causes moisture to permeate into the envelope by way of diffusion and air leakage through intentional or unintentional cracks or other discontinuities in the wall fabric (Lstiburek and Carmody, 1996). This moisture is partially adsorbed and accumulated within the material layers of an assembly. Seasonal moisture cycling in layers causes wood-based materials to alternately expand and contract. Repeated moisture cycling can cause warped and bowed of components, delaminating of glued parts or materials, and the separation of wood members from the structure, causing long term performance problems and compromising durability. Moisture in building materials can be present as vapour, absorbed water, liquid water or ice. Excluding moisture due to rain penetration or convective airflow caused by air pressure or hydraulic differentials, the mechanisms to be modelled are vapour diffusion and liquid transport due to differential capillary suction pressure. Within each material of a building assembly the moisture distribution is governed by the conservation of mass equation. The two-dimensional mass transport equation is:

⎟⎟⎠

⎞⎜⎜⎝

⎛∂

∂+

∂∂

∂∂

+⎟⎠⎞

⎜⎝⎛

∂∂

+∂∂

∂∂

=∂

∂y

MDyTD

yxMD

xTD

xtM

MyTyMxTx (1)

Where: DT and DM = directionally dependent diffusivity coefficients for the temperature and moisture respectively; M = moisture; t = time; T = temperature; and x and y = geometric coordinates. Transport mechanisms due to convection and solar radiation are ignored in Equation (1). The left side of the equation represents the moisture storage within a building material, which can be defined as the ability of that material to absorb or desorb moisture as function of RH and


temperature. As this definition implies, the maximum amount of absorbed moisture that can be stored occurs in an environment with 100% RH, known as the maximum sorption. If more moisture is added to a material already at its maximum sorption condition liquid water will appear in its pores and liquid diffusivity will be the main moisture diffusivity. In wood construction jargon this condition is called the saturation point, i.e. the condition when all voids in the wood structure are completely filled with liquid water. To complete the model for the moisture movement an energy (heat) balance equation is also required. Heat transfer in materials occurs via conduction and convection. Ignoring the vapour and liquid water convective bulk flows, and omitting the heat transfer term for convection, the heat balance equation is:

( )

⎟⎟⎠

⎞⎜⎜⎝

⎛∂∂

∂∂

+⎟⎠⎞

⎜⎝⎛

∂∂

∂∂

=∂

∂

yT

yxT

xtTc

yxp λλ

ρ (2)

Where: cp = heat capacity; λ = thermal conductivity; and ρ = density. Close form solutions of equations (1) and (2) are impossible, except for trivial problems, because of the complex and coupled nature of those equations. Solutions are always non-trivial for domains with irregular geometries, boundary conditions or non-uniform composition. Coupling of the equations is strong in respect of the moisture diffusivity (DM) and temperature diffusivity gradient (DT) which are both functions of moisture content and temperature. To solve equations (1) and (2) numerically requires knowledge of:

• Geometry and composition of the domain (building assembly). The domain is discretized into small easily defined sub-domains known as elements within the finite element scheme (or a grid in the finite difference or finite volume schemes).

• Hygrothermal properties of materials. • Indoor and outdoor boundary conditions, including any surface transfer or symmetry

conditions at the boundaries. • Suitable numerical scheme(s), i.e. stable incremental time steps and numerical control of

parameters.

4.3.2 Domain and boundary conditions The model was developed to analyse the wall panel-to-wall panel joints in walls of the UNB test-building (Sections 4.2.2 and 4.2.3) and other similar constructions. That situation is quite complex and sufficient to yield robust conclusions about model capabilities. Because the model was developed in a two-dimensional domain it can only deal with vertically or horizontally oriented assemblies, but those are very common situations. Figure 15 illustrates the type of situations that can be replicated by such a model(s)


(a) (b)

Figure 15: 2-D domain wall assemblies

As discussed in the previous section (field monitoring activities), the tested wall assembly consists of wood siding, air space, rigid foam insulation, building paper, OSB sheathing, fibre glass insulation, vapour barrier, and plasterboard (Figure 3). Ideally, each wall layer would be treated as separate moisture sensitive materials. However, in practice not all these materials can store moisture, i.e. there are storage and non-storage building materials. Numerical models based on the governing Equations (1) and (2) have provision to handle each material as storage (wood siding, OSB sheathing, fibre glass insulation, and plasterboard) or non- storage (air space, rigid foam insulation, building paper, and vapour barrier) layers. Incorporating storage or non-storage layers in the numerical model enters into definition of boundary conditions applicable to transferring processes for heat and moisture between layers. At the boundary of the multilayer wall, the convective (flow) heat transfer from the surrounding air plus the latent heat from the adsorbed or desorbed water vapour is equal to heat conduction into the surface. For a wall oriented vertically (Figure 15a) this condition gives:

( ) mLTTh

xT

heateff +−−=∂∂

∞λ (3)

Where: hheat = surface heat transfer coefficient; L = latent heat vaporization; m = moisture flux; T = temperature at the building surface to be calculated (in this case of either the wood siding or

Top B.C

….

Top plate

Sill plate

IndoorB.C.

OutdoorB.C.

x

y

A

Bottom B.C

Top B.C

….

Top plate

Sill plate

IndoorB.C.

OutdoorB.C.

x

y

A

Bottom B.C

y

x

studs

Outdoor B.C.

Indoor B.C.

Cross-section A

y

x

studs

Outdoor B.C.

Indoor B.C.

Cross-section A


plasterboard surface); and T∞ = ambient temperature (known inputs for outdoor or indoor climates).

It should be noted that these boundary conditions are applied uniformly along the wall height and thus the so-called stack effect that is known to raise temperatures as function of pressure and building height is not included in the model. At a boundary the moisture transferred through the surface is equal to moisture transferred into the surface. For flux or flow entering perpendicular to the building surface the governing condition is:

)(,, ∞−−=∂

∂+

∂∂ PPS

xMD

xTD MeffTeff (4)

Where: P and P∞ = surface and ambient pressures respectively; and S = moisture permeance or conductance, Analogous boundary condition systems can be applied at the top and bottom of walls, with the applicable flow being in the vertical direction. Another condition that needs to be applied is the transfer mechanisms between each wall component. In the model created it is assumed that when heat is transferred at an interface between two storage layers the temperature is continuous, and if there is moisture transfer the relative humidity is continuous. In a non-storage layer the possibility of any storage of heat or moisture is neglected and the transfers are assumed to be steady. At the interface between a non-storage layer and a storage layer the heat transfer through the non-storage layer plus the latent heat from adsorbed or desorbed water vapour is equal to the heat conduction in the storage layer. This yields the interface condition:

( ) mLTTh

xT

nnsneff +−−=∂∂

−− 1λ (5)

Where: subscripts n denotes the layer index.

For the moisture the diffusion transfer through a non-storage layer is equal to the moisture transfer into the storage layer yielding the condition:

)( 1,, −− −−=∂

∂+

∂∂

nnsnMeffTeff PPSx

MDxTD (6)

4.3.3 Material properties Heat transfer properties: Thermal conductivity, density, and specific heat for the wall components were taken from literature (e.g. ASHRAE, 1997; Kumaran, 2006; Richards et al, 1992) and are summarized in


Table 2. The thermal resistances of building paper and vapour barriers were neglected because they had very small thicknesses relative to other wall layers.

Table 2: Heat transfer properties

Density Thermal Conductivity Specific heat capacity (kg/m3) (Watt/mºC) (J/kgºC)

Wood siding 360 0.11 1630 Air* 1.2 0.03 1000 Rigid insulation* 80 0.043 1050 OSB sheathing 640 0.12 1400 Fibreglass insulation 16 0.048 1050 Plasterboard 670 0.16 1100 Wood stud (pine) 460 0.12 1630 Note: *treated as non-storage layer. Diffusion properties: Moisture diffusivity (DM) in the Equation (1) includes water-vapour diffusivity and liquid water diffusivity caused by capillary transfer when a continuous path of liquid exists within porous material. A continuous path of liquid exists when m exceeds maximum sorption. Diffusivity coefficients, which are functions of vapour pressure, temperature, and moisture and RH have the definitions (Burch and Thomas, 1991):

HMP

Ds

vvM

∂∂

=ρ

µ, (7)

υρ

κρ

s

cw

lMM

PD ∂

∂−=, (8)

s

v

TT

PHD

ρ

µ ∂∂

= (9)

Where: H = relative humidity, P = pressure (vapour/liquid form), к = unsaturated liquid permeability, and µ = permeability, ρs = density of the material. In the analysis, the density of water (ρw) was taken to be 1000 kg/m3, and the viscosity of water (v) was taken as 7.25x10-4 Pa-second. The permeability values, which are function of the relative humidity, were taken from Richards et al, (1992).


Sorption isotherms: Because one of the main input parameters in the model is relative humidity (indoor and outdoor), there is a need to define a relationship between moisture and RH properties for various building components, i.e. sorption isotherms. At operating temperatures of interest temperature does not contribute significantly to isotherm relationships. The sorption isotherm function (Figure 16) can be determined by laboratory measurements and fitting data to the relationship:

)1)(1( 32

1

HAHAHAM

−+= (10)

Where: A1, A2, and A3 = empirical constants determined based on regression analysis of moisture and RH data.

Figure 16: Typical sorption isotherm curve for building materials (Richards et al, (1992)) Detail experimental procedures to determine sorption isotherm for various building materials can be seen in ASHRAE’s paper by Richards et al, (1992). Table 3 presents the empirical constants for the sorption isotherm function used in this project.

Table 3: Constants for sorption isotherm function (ASHRAE, 1997; Richard, et al, 1997)

A1 A2 A3

Wood siding 0.194 2.095 0.769 Air* Rigid insulation* OSB sheathing 0.344 6.177 0.828 Fibreglass insulation 0.263 72.4 0.843 Plasterboard 0.025 9.075 0.935 Wood stud (pine) 0.212 3.43 0.811 Note: *treated as non-storage layer.

Relative humidity, H (% )

Moi

stur

e co

nten

t, M

(%)

10

20

30

20 40 60 80 100

)1)(1( 32

1

HAHAHAM

−+=

00

M aximum sorption

Relative humidity, H (% )

Moi

stur

e co

nten

t, M

(%)

10

20

30

20 40 60 80 100

)1)(1( 32

1

HAHAHAM

−+=

00

M aximum sorption


4.3.4 Numerical solution procedure The finite element method was used to discretize the mass (moisture) diffusion and heat transfer equations. An overall system of equations, based on Equations (1) and (2), applying to the modelled domain was solved to determine the distribution of moisture (M) and temperature (T). Details of the finite element formulation of heat and moisture transfer including and the algorithm are given in Appendix D. Computational software MATLAB was used to develop a computer coding for solving the system of equations including providing pre-processors (meshing, boundary conditions) and post-processors (plotting). In the finite element method, the accuracy of the numerical solution for time-dependent problems is a function of the combination of element type, meshing strategy (element sizes) and time-marching scheme (increment lengths). For a two-dimensional model, element size parallel to the heat flow direction across the wall was made smaller than that perpendicular to the flow direction (see, for example, Figure 20). Linear four-point elements were used. For unsteady state analyses, the one-year analysis period was divided into increment of 1200 seconds (20 minutes) which was necessary to obtain stabile numerical solutions. Depending on the specific meshing involved, for a two-dimensional model the total time running for a one-year simulation was between 12 and 24 hours. 4.3.5 Model verification One-dimensional model with one building material: As a first step in the model verification steady state heat and moisture transfer across a uniform wood layer was analyzed. The wood thickness was 420 mm and meshed with 1 mm linear elements. The outdoor environmental conditions were set as –3ºC and 48% RH; and the indoor environmental conditions were 27ºC and 88% RH. Initial moisture content was set at 10% and the initial temperature at 27ºC. The temperature and moisture profile across the thickness after one year was predicted to be as shown in Figure 17. That figure also shows, for comparison, the corresponding solution obtained with a finite different method solution. Results are in quite good overall agreement between finite element and finite difference schemes, with a tendency for result about 5% higher than the result by the one-dimensional finite difference model.


(a) Temperature distribution

(b) moisture distribution

Figure 17: Finite element and finite difference comparisons One dimensional steady-state multi-layer wall assembly: The finite element model was then used to analyze moisture distribution in the wall assembly of the UNB test building. Detailed dimensions of wall layers are as shown in Figure 3. To simplify the domain of analysis, the wood siding was modelled as having a uniform thickness of 10 mm. The wood siding, OSB and plasterboard layers were treated as storage materials, while air space, rigid foam insulation, building paper and fibreglass layers were treated as non-storage materials. Outdoor conditions were initially set at 9ºC and 78% RH, and indoor conditions at 27ºC and 50% RH, with an initial temperature of 27ºC and initial RH 65% for all storage layers. The predicted changes in average RH after about four months are shown in Figure 18. The final temperature


and moisture content distributions across the wall, after one year, are shown in Figure 19. As Figure 18 reveals, the moisture movement in the wood siding is predicted to be a very slow process, taking 6 weeks for the wood siding to increase average RH from 65% to 70%. At the final state, the average RH was 72.5% in wood siding, 62.6% in OSB and 51.5% in plasterboard. As seen from Figure 19, RH in OSB is not predicted to be strongly affected by the environmental conditions. This suggests that wood siding and vapour barriers are effective obstacles to movement of moisture across a wall. Compared with the moisture movement, the heat conduction in the storage layers was rapid. As is expected, the non-storage layers including rigid foam and fibreglass decrease the heat loss across the wall. From these results the finite element model gives reasonable qualitative answers concerning processes heat and moisture transfer for simple multi-layer wall assemblies. No quantitative comparison of one-dimensional prediction with test data was performed because that was expected to be inaccurate based on reports in the literature.

Figure 18: Moisture content in storage layers

(a) temperature (ºK=ºC+273) (b) moisture (%)

Figure 19: Predicted temperature and moisture distributions across the wall

Rel

ativ

e hu

mid

ity (%

)


Two-dimensional unsteady-state analysis: The next step is to verify the two-dimensional finite element model using the actual field monitoring results. Actual dimensions of the wall layers and actual hourly weather conditions were used as model inputs. The wall cross section was oriented horizontally in the numerical representation to permit incorporation of the wall panel-to-wall panel construction joint (joint), Figure 20. This cross section illustrated corresponds to the middle-height location in the real joint. A very small airspace (non-storage component) was modelled in the butt-joint interface to allow air movement (air leakage). So far calculation of air leakage values is not included in the model, but that is possible and will be done later for purposes of estimating energy efficiency of construction joints. The indoor temperature condition applied is that shows in Figure 7, whereas the indoor RH was taken as constant at 50% during one year of analysis. The outdoor air temperature and RH conditions are as in Figures 9 and 11 respectively. All hygrothermal properties discussed in the previous section were used in this analysis. In total 495 node points and 432 four-node elements were used in the model. Applying stable time step ∆t = 600 seconds, it took about more than 12 hours to run a one year analysis using the computational platform MATLAB. Numerical instability, which was indicated by non-converging parameters (temperature and moisture), occurred when the total time analysis was extended beyond one year. This problem can be eliminated by reducing the length of time steps and/or using a fine meshing strategy, but that can lead to much longer computer running times.

Figure 20: Geometry and FE mesh of the wall assembly To compare the model with the field monitoring results, the temperature and RH corresponding to locations of the sensors installed in the wall were plotted during for the one year study period. Here it is chosen to show data for the thermocouple labelled Channel 21 and the RH sensor labelled Channel 20, Figure 21. As can be seen, the predicted temperature distribution is in close agreement with test results. Channel 21 readings are given in Appendix A. It can be seen that the

outdoor

indoor

butt-joint

wood siding

OSB sheathing

fiberglass insulation

gypsum board

studs

air space & rigid insulation

T & RH sensors outdoor

indoor

butt-joint

wood siding

OSB sheathing

fiberglass insulation

gypsum board

studs

air space & rigid insulation

T & RH sensors

plasterboard (gypsumboard)


temperature ranges for both the model and tests are between about 0 and +25ºC. There is a small discrepancy in predicting maximum temperature during summer. The model predicted maximum temperature ranges between June and August, while the test results were that the period was July to September. It is speculated that this is due to addition of wood siding to the test building during the observation period. Such a construction modification would obviously alter the response. (Note: the simulation was performed assuming the complete wall system throughout). The relative humidity results indicate a fair agreement with the test result. As shown in Figure 21 and Appendix B (Channel 20), the predicted relative humidity range is 55 to 67%, while the test range is 40 to 70%. The curve trend indicates that the model is in agreement with the test result with small discrepancy found during the early months of the recording period. The discrepancy between the model and test could be due to various factors, apart from construction activities conducted during the monitoring period. Air leakage as well as convective heat and moisture movement could be major factors in this discrepancy. However, the overall trend of the relative humidity pattern after about half of the total study period indicates that the model can be used with confidence to predict the performance of alternative construction joint configurations. This is an especially valid deduction if the model is used to assess likely relative performances of alternative designs.

Figure 21: Predicted temperature and relative humidity for one-year period of the sensors shown in Figure 20

Rel

ativ

e hu

mid

ity (%

)


Figure 22 presents the temperature and RH distributions at selected times. From the temperature distributions, it can be seen that a high temperature gradient occurs at the wall joint. This is because the wall studs either side of the joint have higher thermal conductivity than the insulation. This leads to high heat flow and energy losses at that location. High moisture accumulations are predicted in the wood siding. If the surface layer is not sealed to prevent moisture exchange That is not however a major concern because its temperature always follows the outdoor temperature pattern resulting in a balance between wetting and drying. OSB sheathing and stud are the most sensitive parts with respect to excessive moisture accumulation issue. As shown in Figure 22 moisture accumulations at the OSB sheathing and the two-stud joint at the end of one-year period is between 10-12%, which was in accordance with moisture measurement taken at that time.

time = 2 hours time = 83 days

time = 278 days time = 365 days

(a) temperature contour history (ºC)

time = 2 hours time = 83 days


time = 278 days time = 365 days

(b) moisture content history (%)

Figure 22: Temperature and moisture histories near the wall panel-to-wall panel joint 4.3.6 Prefab wall panel-to-wall panel joint analysis As discussed in detail in Section 2, building with prefab wall panels must contain discrete construction joints. Such joints are points of articulation and will experience deformations due to externally applied structural and environmental loads, and they must be designed properly to accommodate that without loss of functionality. Apart from structural integrity requirements, joint must be designed against air leakage and creation of temperature bridges. In this context, the need to prevent or at least mitigate air leakage should not be confused with creation of conditions of excessive humidity within buildings as has been experienced with some modern ‘tight construction systems’. Here it is presumed that prevention of excessive leakage through wall should go hand in hand with provision of an efficient and automated HVAC system, or at minimum an adequate automated air exchange system. Based on this the following summarises an analytical investigation of alternative wall panel-to-wall panel joint system designs using the numerical model created during this project. There are many types of systems that are or could be used, such as butt, lap and tongue-and-groove (t&g), Figure 23. Analysis of a butt joint, and a tongue-and-groove joint with triple studs (Figure 24) is presented. In the butt-joint system airflow was assumed to be in a straight line starting from the OSB to just before the plasterboard layer. In the t&g system airflow was assumed to take a longer path as shown in Figure 24, and the finite element mesh used for the tongue-and-groove system was modified to facilitate that flow path. Airflow with a rate 1 unit volume/hr was applied to these paths. That is not the actual airflow recorded in the test building but is an assumed value that can sensibly be in comparative analysis. To have consistent numerical calculation, the mesh size used was similar for butt-joint and t&g systems. In total 550 nodal points and 486 four-node elements were used for the t&g mesh. All boundary conditions applied were the same as before (Section 4.3.5). Total time interval for the analysis was one year, with time steps of 600 seconds.


(a) butt-joint (b) lap-joint or

(c) tongue-and-groove

Figure 23: Types of wall panel-to-wall panel joints Figure 24 shows the predicted moisture distributions after one year of service. Comparing the output for the two systems, it can be concluded that the t&g joint produces lower moisture contents, particularly in the studs adjacent to the joint (10 to 15%). In the butt-joint system the corresponding moisture content can reach up to 24%. This is because the t&g joint dries much more quickly than the butt-joint system. Small numerical instability occurred during the computation, as indicated by localized oscillation in the moisture distribution values around the wood siding layer. This is not a major issue and can be corrected via alteration of the numerical strategy, and as will be done in future computations adopting smaller time steps and a finer finite element mesh.

shorter flow path longer flow paths

butt-joint type t&g joint


(a) shorter flow path (b) longer flow paths

Figure 24: Comparison of moisture distribution between two panel joint systems 4.3.7 Finding summary of the numerical modelling

The finite element model of heat and moisture flows developed in this project was used to investigate hygrothermal performance of a particular light-frame building with prefab wood wall panels. It was found that the numerical simulation provides proper airflow path, temperature and moisture content predictions, including at the wall panel-to-wall panel construction joints that are integral to the type of prefab system. Using the verified model it is demonstrated here that it is possible to reengineer the construction joints in a manner that reduced moisture depositions around those joints without sacrificing either energy efficiency or structural performance. Discussion of structural performance of panellized wood joints is beyond the scope of this project but has been investigated in the previous NRCan funded project “FCC25 Connection Systems for Prefabricated Wall Panels” (Mohammad, 2006).


PART II

Handling and transportation aspect of modular wood buildings


5. Handling and transportation aspects of modular wood buildings 5.1 Background Quality of housing is a function of performance at every stage of construction and in service. Handling and transportation of building modules and manufactured homes is an important part of the overall life of modular houses because for most units that is when they will experience the highest mechanical loads. In North America virtually all the transportation is by road. The common transportation system is a wooden building module sitting on the top of a standard flatbed-trailer pulled by a tractor unit (tractor-trailer). Some provinces of Canada and states of the USA impose strict control on the maximum size of loads trailers can carry, and that dictates the possible size of house modules. Typically a trailer can carry a ‘regular module’ 13½ feet (4.11 m) wide unit with an average weight of about 15 or 16 tons (about 150 kN). Some regulations allow larger units up to 16 feet (4.88 m) wide. Modules must be designed with adequate resistance to structural stresses and deflection arising from handling and trucking. To accommodate this, some manufacturers reportedly use up to 30 percent more material than would be required with on site, stick building, construction (PATH Inventory, 2003). From what the project team have determined, the forces developed on modules during transportation have not been quantified before and the amount and placement of the extra reinforcing materials is based on judgment rather than objective analysis. There is little to suggest that current factory and site handling practices or highway transportation practices often cause damage that will impair structural performance of complete buildings. However, even minor damage that disrupts continuity in the envelope can significantly affect air leakage and moisture deposition characteristic of the building in service. From a structural point of view, inappropriate placement of reinforcing materials will concentrate stresses at vulnerable locations and make modules or whole manufactured homes susceptible to cracking in internal or internal finishing materials. Damage induced leakage will affect air exchange and heating or cooling characteristics in the short term and durability in the longer term. There is need for special design and construction methods and technologies and/or optimization of transportation practices, to avoid or at least mitigate possibilities of handling or highway transportation induced damages. Structural stresses developed during transportation are somewhat different from those developed during the service life of a house. Damage will occur if stress anywhere exceeds the strength of deformation that a component (structural or non-structural) can withstand. Transportation stresses occur due to wind and road induced vibration, braking and centrifugal forces. For example, wind pressures will be very high if wind blows in the opposite direction to that in which a truck is moving. Vibrations due to road roughness (impacts) and wind turbulence as trucks pass each other, accelerate or decelerate causes high stresses especially if vibration induces resonance. Resonance occurs when the natural frequencies of the transportation forces match one or more natural frequencies of the vehicle or the module/unit seated on it. There is also potential for the so-called ‘beating effect’ when the


difference between two forcing frequencies matches the difference between two response frequencies. 5.2 Related Work Despite the large volume of modular units and manufactured homes being transported by road in North America, there has been relatively little study on the topic of wind and vibration forces. Detailed study of lateral forces due to wind on manufactured homes was conducted at Colorado State University (Creighton, 1997) and the University of Wyoming (Schmidt et al, 2000). The study by Creighton focused on numerical modelling of complete homes using finite element analysis. The major load input was wind load obtained from the literature for pressures on closed transport trailers. No testing was performed to determine wind forces or the strength of a completed unit. The modelling was “verified” via comparison with test results collected under laboratory conditions for walls, truss and floor subsystems. Thus, there was in fact no verification of the modelling of the interfaces between subsystems of houses or completed systems. This is potentially a major weakness as detailing of interfaces can be critical to adequacy of models for complete systems. Creighton’s conclusion was that his numerical model could be applied to types of manufactured housing other than he investigated. He recommended performing full-scale testing under real wind load conditions. Schmidt et al (2000) conducted a comprehensive study on a typical ‘singlewide’ manufactured home under wind load using field-testing and numerical modelling. Key components and connections were also tested to provide input data for finite element analysis. The field measurement results were used to validate the analytical model. Schmidt et al recommended continuing the work with the focus on comparison between wind tunnel and the field studies. The two studies mentioned above took no consideration of road related forces such as those due to vibration. Road roughness contributes significantly to vibration forces. The investigation by Marcondes and Singh (1992) on the suitability and use of road roughness index data to estimate levels of vibrations in transported vehicles is an alternative to using the direct measurement of acceleration levels. Pavement elevations were classified based on the International Roughness Index (IRI) converted into an observed PSD (Power Spectral Density) acceleration at the truck bed. Marcondes and Singh concluded that IRI can be used to predict vertical acceleration in vehicles, with the approach most accurate at frequencies of vibrations between 5 and 12 Hz. IRI is a good indicator of ride quality for heavily loaded tractor-trailer systems and shows good results at 10 Hz. Spectral analyses is based on an averaging process that smears the effects of discrete potentially damaging events like impacts on potholes or during crossing of rail tracks. Thus IRI appears inappropriate for assessing likelihood of damage to prefabricated building modules and manufactured homes. Singh et al (2006) measured and analyzed the vibration levels in commercial trucks and the effect on packaged goods. It was found that increasing truck speed increased vibration levels and thereby caused more damage to packaged fruit. Laterite (unpaved gravel surfaced) road conditions produced the highest vibration level followed by concrete highway and asphalt road conditions. The damage to packaged produce was greatest in the uppermost parts of the


containers for every combination of road, truck type, and travelling speed, which corresponded to the highest vibration levels recorded. 5.3 Field test program 5.3.1 Parameters measured In coordination with a local modular housing company, field measurements were devised for forces and deformations on a modular housing unit during in-factory lifting and highway transportation. Those field results were used as input for verifying numerical models for lifting and highway transportation responses of prefabricated units. Parameters measured during transportation were: (1) wind speed and direction with respect to the truck axis; (2) wind forces on surfaces of modules; (3) deformations in modules; and (4) forces due to road roughness, cornering and acceleration and deceleration of the truck. During lifting/handling processes at the manufacturing facility static deformations around building openings were measured because practical experience is that wall openings are loci for damage. The measurement system was assessed in trials in a controlled (laboratory) environment with a quarter scale wood building before the fully realistic application. A reduced scale model was placed on a small flat bed trailer pulled behind a light truck for road testing. Pressure sensors and accelerometers were installed and the data acquisition system installed. This proved that the measurement system was functioning properly. 5.3.2 Description of the prefab home tested The prefab home chosen for the study was the longest type transported on North American roads, 74 ft. (22.6 m), Figure 25. This is a typical three-bedroom mini home having a large living room, and large door and window openings. Figure 25 (b) shows the plan layout of the building. Such units are a type prone to damage during handling and transportation.

(a) 16’x74’ prefab mini home


(b) Plan view

(c) During lifting

(c) During transportation

Figure 25: Prefab mini home tested

5.3.3 Field instrumentation All instrumentation was installed when the mini home was at the end of the factory production line. Eleven deformation sensors were installed in the large living room area where several large openings (doors and windows) are located. Structurally the living room represents the weakest


section in the mini home because there are no internal walls (partitions) and the external wall has openings. It was anticipated that any damage would occur mostly in that area. Six of the deformation sensors were located at the top corners of doors and windows using highly sensitive clip gauges, and five sensors were placed at wall-to-roof junctions using LVDTs (Linear Variable Differential Transformers). The high sensitivity clip gauges were placed on the plasterboard (drywall) surface oriented at ±45º

relative to the horizontal direction. This facilitated

monitoring of any crack development from the corners of wall openings because experience has shown that cracks propagate at about ±45º to the horizontal, Figure 26. Eighteen pressure taps were installed on exterior wall surfaces: six of them on each of the long side elevations, four on the front elevation (face near the truck cab), and two on the rear elevation. The pressure taps measured differential wind pressures applied to walls (external minus internal pressure). Holes for pressure taps were drilled through the wall studs to avoid damaging insulation material. No pressure taps were installed on the roof surface because it would have caused damage unacceptable to the owner. It is anticipated that, in the future, pressure forces acting on the roof surfaces will be estimated using computer modelling and/or wind tunnel simulations, and possibly further field tests. Sensors locations are shown in Figure 26.


End Elevations

Figure 26: Sensor placement (A&B=deformation, P= pressure taps)

An anemometer to determine wind speed and direction was mounted on the front side of the building just below eaves level. The ideal scenario for the anemometer’s location would be in the middle of the building mounted a few meters above the top of the roof to avoid any air turbulence. However, this could not be implemented due to restriction from the department of transportation on the allowable height of transported objects. The dimensions of the mini home itself as well as truck’s size already represents the allowable limit for an object transported on public roads. All of the sensors were connected to high-speed data logging equipment with sophisticated data acquisition software. Two different data acquisition systems (DAS) were used. The first DAS captured low frequency signals from the anemometer (measuring wind speed and direction), pressure taps (differential wind pressures on module walls), and deformations in the module. The second DAS captured high frequency signals recording accelerations at various locations on the module and truck. Four acceleration sensors were connected the second data logger and mounted in the vibration prone areas of the building, Figure 25. Two 500-Watt battery packs were used to supply power for the data loggers and computers during transportation. A specially designed wooden cage was used to minimize the effect of vibration to the computer data acquisition systems as shown in Figure 27.

Figure 27: Computer data acquisition system Placement of sensors, and reasons for the choice of their locations, are as follows. Judgment for pressure taps locations was based on ‘Gaussian points’ for the wall surfaces, because as is well known from numerical analysis knowledge of values taken by variables at the Gaussian point’s permits accurate integration of their functions over the surface. In this case the functions defined


wind pressures. Some errors undoubtedly occurred because of air turbulence. Nevertheless installing pressure taps at the Gaussian points yielded reasonable accuracy of total forces and an approximate indication of the distribution. In this project, it was decided to use six Gaussian points on side-wall surfaces and four points on end-walls. Selection of the pressure tap locations was also based on computational fluid dynamic (CFD) analysis of a prefab mini home during transportation to determine apparent wind pressures (Appendix E). It was concluded that under uniform airflow relative to the truck the wind pressure distribution is quite regular. In practice, the apparent wind could come from many directions resulting in quite irregular pressure distributions. Determination of acceleration sensor locations was based on the sensitivity of those locations to the structural vibrations, such as where the transported mini home overhung the flatbed-trailer. Selection of the deformation sensor locations was based on structural analysis of the mini home and practical experience. 5.3.4 Field test procedure The field test was divided into four major stages:

(1) lifting of the mini home at the factory, (2) short transportation from the factory to a holding area (temporary storage) in the factory

yard, (3) lifting of the mini home to a trailer in the factory yard, and (4) road test.

The mini home was lifted in stages (1) to (3) using six hydraulic jacks positioned as shown in Figure 25(c). These hydraulic jacks were operated at nominally the same speed, with the total time needed to lift the mini home to a truck taking about 3 minutes at the factory and 15 minutes at the factory yard (holding area). The difference in the time frame was because there are two steps in the factory yard lifting, i.e. lifting the module from a temporary trailer and lowering the module to temporary supports. Additional time was also needed to level the mini home’s position on the temporary supports, because the area is unpaved and not level. Deformation and acceleration histories were recorded during all stages of field tests. Wind pressure and wind speed and direction were only recorded during the road testing of stage (4). Deformation, wind pressure, and wind speed and direction data were recorded at one-minute intervals. Acceleration data was recorded at 200 Hertz (200 hundred data records every second). This was because most of the vibration included shock (impact) phenomena lasting for relatively sort time periods. A special computer program was designed for the data acquisition system to manage and store the huge amount of acceleration data. The typical mini home configuration during transportation can be seen in Figure 25 (c). The road test was especially scheduled for purposes of the project. Under the coordination of the local transportation authority the transportation route represented both major highway and rural roads typical of Maritime Canada. During the road test, two stops were made to check the data acquisition systems including the crucial power supply, and to record the development of visual damage (cracks). In total the road test took about two hours.


5.3.5 Field test results Damage observation: Visual methods were adopted to assess damage. Observation of damage indicated that major cracks occurred in plasterboard around building openings such as windows, doors, and in the ceiling. Figure 28a shows an example of a crack initiating at the corner of a door opening. Significant longitudinal and transverse cracks were also observed at ceiling opening and at wall to ceiling-roof junctions in plasterboard (Figures 28 b& c). As expected, these damages occurred mostly in the middle section of the mini home in the spacious living area. These damages were recorded in detail at end of each stage of handling and transportation. It was observed that lifting practices adopted by the collaborating prefab company are the major cause of damage in large mini homes. Damage caused by that lifting was observed to increase due to vibration and wind pressure induced cyclic fatigue processes during road transportation. Deformation analysis: The corners of window opening and junctions between the ceilings and walls are susceptible to large deformation, as confirmed by visual observations of relatively major cracking at such locations. No significant displacements occurred during lifting operations inside the factory. Maximum displacement was observed during lifting of the mini home in the factory yard. The main cause of the damage was the unevenness of the jacking system placed on rough ground in the factory holding area. That caused reaction forces to be irregular and to develop critical levels of localized stress. Table 4 shows peak deformations recorded during all the processes. In general it can be seen that the recorded deformations were accentuated from process stages (1) to (4). Looking at sensors located at the corners of door/window openings (e.g. sensors A3 and A4, A10 and A11), it can be noted that the wall was subjected to reversed deformations (and therefore reversed stresses), i.e. one sensor was under compression (e.g. A3&A10) and another was under tension (e.g. A4&A11). This indicates that concentrated racking and/or shearing forces were induced in the walls near the opening leading to crack initiation at one corner (Figure 28a). During factory-yard lifting deformations at B3 and B6 show relatively large differential values (deformation at B3 about ten times that at B6). This indicates that the building was subjected to forces twisting it about its longitudinal axis. Significant damage was found in the skylight area where the plasterboard could not resist the imposed deformation/stress (Figure 28b).

Table 4: Peak deformations at various process stages (mm)

Notes: - (1) = factory lifting; (2) = short transportation; (3) = factory-yard lifting; (4) = road test - Notation for the deformation sensors and their location are shown in Figure 26

Process Stage

A1 A2 A3 A4 A10 A11 B1 B2 B3 B4 B6

1 0.0022 0.0035 0.0042 -0.004 0.003 -0.0007 0.0049 -0.008 0.015 0.065 0.015 2 0.015 -0.015 -0.038 -0.013 -0.14 0.105 -0.011 -0.02 0.015 0.015 0.02 3 0.12 0.13 -0.20 0.1 0.84 1.43 -0.01 0.04 0.13 0.12 0.01 4 1.0 -0.65 -1.0 -0.04 -0.7 -0.42 -0.014 0.048 -0.15 0.24 0.03


Figure 29 shows an example of a deformation history recorded at sensor A10, the location of crack is shown in Figure 28a. From the figure it can be seen that the deformation initially was oscillating about zero and shifted the deformation at equilibrium during course of operations. When deformations oscillated about zero no damage was observed at the corner, but as the reference position changed small cracking was initiated. Complete plots of the deformation histories during the whole processes (1 - 4) are shown in Appendix F. During short transportation from the factory to factory-yard, sensor number A10 initially exhibited relatively small deformation. Then during about the middle third of the process larger deformation occurred and reached 0.13 mm, with rapid changes that indicate some kind of impact process. This was attributed to large vibration-induced deformation due to the rough ground of the factory yard. In the remaining time of process (2) the reference deformation point shifted away from zero indicating that cracking was propagating and causing unrecoverable deformations. During the factory lifting, sensor (A10) showed three distinct behaviours. Initially during about the first 280 seconds deformations were almost zero. Then a constant deformation of -0.12 mm was observed, followed by a sudden jump to -0.4 mm, before it changed suddenly to a large positive deformation that reached +0.8 mm. It should be noted that small hairline cracking had already developed during factory lifting operations. Finally, during the road test the deformations recorded by sensor A10 shifted from positive to negative with highly fluctuating values. This could indicate that damage was accentuated during the road test.

(a) corner of door opening (b) ceiling opening (skylight)

(c) roof-to-ceiling junction

Figure 28: Damages observed


Figure 29: Deformation around a window corner (sensor A10) during factory lifting

Wind speed, direction and pressure: Wind speeds recorded were apparent values that are a superposition of the truck speed and the environmental wind speed at given times. The maximum wind speed during transportation approached 200 km/hr, with average value 100 km/hr (Appendix G). This is equivalent to what might be expected during a Category 3 Hurricane according to the Saffir-Simpson Scale. Wind directions recorded fluctuated highly ranging from 0º

to 130º

with average of 54º

with respect to

the longitudinal truck axis. (Note: The direction is measured clockwise relative to the mini home axis looking from the back of the flatbed trailer towards the front of it). Obviously wind directions depended on the weather conditions, the route and local instantaneous topography, and could have differed substantially for another route or the same route at another time. The key consideration is that the data is adequate for verification of models that can be applied to gain generalized understanding of what range of wind pressures modules might experience during transportation. In general, wind pressures recorded were relatively small due to the angles of wind incidence that occurred between the building surfaces and the environmental wind direction. Also, because of these angles alternated most building surfaces experiences fluctuations between positive and negative (suction) pressures. For the particular journey that was studied pressure data was within the range of values expected based on a preliminary computational fluid dynamic (CFD) model of the mini home and truck (Appendix E). On the front face of the mini home, with four pressure taps, the pressures at those locations indicate higher that values are typical at lower than upper pressure tap locations. On side faces (walls parallel to the axis of the building) were there were six pressure taps per face, wind pressure values fluctuate between –50 to +50 Pa. These values are also within the range of CFD model predictions. However, there is a disagreement between the CFD model and the pressure pattern recorded on the rear face of the mini home. This is


probably due to strong turbulence at that location resulting from the real arrangement and environmental factors being more complex than in the model. Future work will compare the wind pressure distributions deduced from the test with a wind tunnel simulation, as well as a more fully developed CFD model. The full-scale tests demonstrate the importance of considering actual pressures on building surfaces rather than simply deducing what they might be from wind speed measurements and pressure coefficient information applicable to completed buildings installation on a foundation. Vibration: Vibration response is analyzed through acceleration data recorded during the factory yard lifting (lowering and lifting) and road transportation processes. Acceleration was measured in units of gravitational force (g’s) at the locations shown in Figure 26. Due to a limited number of data channels on the available data acquisition system, accelerations were recorded only in the vertical direction. From field observation, movement of lifting equipment (hydraulic jacks) generates mainly such motions in mini homes as do shocks (impacts) at the end of the lifting process. Acceleration records were transformed into ‘readable parameters’ to assess vibration levels developed during each process stage, to assess whether vibration levels are capable of inducing damage, and to estimate vibration forces developed during transportation. By integrating the acceleration data over time, particle peak velocities (PPV) were determined at building surfaces for locations where accelerometers were placed (Table 5). PPV is a parameter that indicates the likelihood of damage. Based on Splittgerber (1978), damage due to vibration can occur for PPV values ≥3 mm/s (Table 6). The PPV values developed during transportation fall in the range associated with the category ‘minor damage’ to building components. Minor damages observed were cracking in plasterboard linings of the mini home and other superficial damage that did not immediately impair the structural system (Taylor, 1994). The PPV values developed during transportation were much higher than for lifting processes (Table 5). Another important parameter in vibration analysis is Root-Mean-Square Acceleration (RMS-A). RMS-A can potentially be used to assess the energy level developed during a vibration episode. Shock due to impacts in particular should be assessed using RMS-A records. Figures 11 and 12 in Appendix H show impact loads detected during lifting processes. It can be seen in Figure 13 and 14 of that appendix that no substantive impact loads were generated during the road transportation period.


Table 5: PPV and RMS data Locations, Fig. 1 Lowering in the factory

yard Lifting in factory

yard Local roads

Highway

PPV (mm/s) 3.03 6.65 23.64 21.21 C0

RMS-A (g2) 0.003 0.006 0.087 0.10

PPV (mm/s) 8.72 7.87 33.12 17.49 C2

RMS-A (g2) 0.003 0.004 0.071 0.06

PPV (mm/s) 11.09 9.27 50.61 31.28 C3

RMS-A (g2) 0.011 0.010 0.114 0.06

PPV (mm/s) 6.79 6.47 36.79 18.32 C1

RMS-A (g2) 0.005 0.006 0.053 0.05

Table 6: Vibration induced damage level (from Splittgerber 1978) Damage level PPV Threshold 3-5 mm/s Minor 5-30 mm/s Major >100 mm/s

Other data derived from the acceleration histories includes mean and peak accelerations, incidence of impact loads, and natural vibration frequencies of the mini home (Appendix H). These data are used as input for, or verification of, finite element dynamic analysis models of the mini home. 5.4 Numerical modelling 5.4.1 Finite element model Numerical models for prefab homes during lifting and transportation were developed using the finite element (FE) structural analysis software SAP2000 Version 10.1 (CSI, 2006). Three-dimensional (3-D) models were developed incorporating all-important components of the building, Figure 30. Linear orthotropic shell elements were used to model wall and roof sheathing, plasterboard, and floor sheathing. Wall studs members were modelled using orthotropic shell elements instead of common frame elements, because that accommodated investigation of force distributions between sheathing and plasterboard. Floor joists and roof framing were modelled using orthotropic shell elements. The interfaces / connections between framing and sheathing components were modelled using link elements composed of internal springs (axial, shear, and rotational). The finite element mesh formed comprised of 17,332 nodal points, 9,585 frame elements, 9,479 shell elements, and 6,483 link elements.


Figure 30: 3-D finite element model of a prefab mini home

5.4.2 Boundary conditions (B.C.) B.C. during lifting process: The lifting of manufactured homes in the factory is illustrated in Figures 31 and 32a. Typically three electro-hydraulic jacks at both front and rear sides of the home are positioned as per the dimensions shown in Figure 31. Once the jacks are positioned each is switched on to start the lifting. To simulate this condition in the finite element model hinge supports were imposed over a contact area equivalent to where each jack grips, Figure 32b. By modelling lifting supports as hinges, the structure was constrained against moving in the translational directions but allowed to rotate (permitting bending and twisting of the whole unit), which is the typical behaviour observed in practice. In the field process each jack is switched on manually separately and time delays and lack of synchronisation is unavoidable. This usually leads to differential movements of the housing unit at the support positions. This situation was simulated as is reported later in this report.

Figure 31: Lifting position


(a) Typical lifting process on shop floor

(b) Modelling the support condition Figure 32: Modelling boundary supports during lifting

B.C. during transportation: Transporting prefab homes is similar to transporting packaging containers. In packaging containers, a container is usually positioned on the flat bed of a tractor-trailer with relative lateral movements restricted by bolting it to the flat bed. Similarly, a prefab home is usually placed on a flat bed and its lateral movements restricted, but in that instance using chains connected to the bottom part of the floor framing system, Figure 33a. For this process it may be assumed that because of the self-weight of the home relative motion between the home’s floor sub-system and the flatbed of the truck is negligible provided that during transportation the trailer moves at reasonable speeds. During the road testing it was observed that at some locations there were gaps between the floor and flat bed, as shown in Figure 33b. To simulate the actual conditions in the finite element model, hinge constraints were used over the area of contact between the home and the flatbed as shown in Figure 34. The total contact area was 8’6” x 74’ (2.59 m x 22.555 m). No contact between floor beams and the flat bed was assumed in modelling, Figure 35.


(a) (b)

Figure 33: Gaps between floor beams and flat bed

Figure 34 The arrangement of mini home on the flat bed of the tractor-trailer


Figure 35: Modelled support conditions during transportation

5.4.3 Materials and joints/connections models Wood components including studs, OSB sheathing, floor beams, roof trusses, and roof sheathing were modelled as linear elastic orthotropic materials. Plasterboard which is composed mostly of gypsum was also assumed to behave as a linear elastic isotropic material (Cramer et al 2003). Mechanical properties were taken from related studies at UNB and literature. Wood component properties were taken from Mi (2005) and Winkel (2006), and plasterboard properties from the Gypsum Board Association (2005). The mechanical properties used in the model are summarized in Table 7. The behaviour of sub-building connections/joints (wall-to-wall, wall-to-floor, and wall-to-roof) was modelled using link elements in SAP2000 library (Mi, 2005). Link elements permit transfer of moment and resultant forces in any direction. Connections between each wall component (OSB-to-studs and plasterboard-to-studs) were also modelled using link elements, Figure 36. Since factory practice is to glue connections in addition to nailing or screwing them, rigid behaviour of link elements was assumed. The behaviour of sub-building connections/joints (wall-to-wall, wall-to-floor, and wall-to-roof) was modelled using link elements in SAP2000 library (Mi, 2005). Link elements permit transfer of moment and resultant forces in any direction. Connections between each wall component

Structural hinges


(OSB-to-studs and plasterboard-to-studs) were also modelled using link elements, Figure 36. Since factory practice is to glue connections in addition to nailing or screwing them, rigid behaviour of link elements was assumed.

Table 7: Properties used for the analysis

Figure 36: Connection details modelled using link elements

Modulus of elasticity (MPa)

Poisons ratio Shear modulus (MPa)

Element Element type used (thickness in mm)

Ex (Dir 1)

Ey (Dir 2)

Ez (Dir 3)

Plane 12

Plane 13

Plane 23

Plane 12

Plane 13

Plane 23

38x140 stud Shell (38) 12000 12000 12000 0.3 0.3 0.3 3650 3650 3650

Plasterboard Shell (12.7)

13000 13000 13000 0.3 0.3 0.3 5540 5540 5540

OSB sheathing

Shell (11.1)

3000 5000 3000 0.3 0.3 0.15 1200 1700 1200

Floor sheathing

Shell (15.9)

3000 5000 3000 0.3 0.3 0.15 1200 1700 1200

38x235 Floor joists

Shell (38) 12000 12000 12000 0.3 0.3 0.3 3650 3650 3650

38X89 roof truss

Frame 12000 12000 12000 0.3 0.3 0.3 3650 3650 3650

Roof sheathing

Shell (9.5) 3000 5000 3000 0.3 0.3 0.15 1200 1700 1200

Node j

Node ik, connection stiffness

Link element model

Node j

Node ik, connection stiffness

Link element model


5.4.4 Loads During lifting the prefab unit was subjected to the forces due to its self-weight, which was around 160 kN. The analysis of lifting processes was assumed as static and with loads acting downward. No superimposed dead loads due to finishing materials and fixtures like cabinets were included in the analysis. During transportation loads applied included road roughness induced vibration forces and wind pressure forces. To simulate the conditions of vibration forces maximum and minimum acceleration data collected during the road test were input. The movement of mini homes in the lateral and longitudinal directions due to sudden accelerations due to braking, fast turns and fast starts were modelled as equivalent to lateral floor movements as occurs during earthquakes. Because no lateral accelerations were recorded during the field test numerical values for dynamic lateral loading were applied using earthquake data available in the SAP2000 library. This is only a rough equivalence but is sufficient to indicate whether such motions are potentially damaging. An earthquake with a magnitude of around 7 on the Richter-Scale was used, which is consistent with provisions of the US Uniform Building Code (CSI, 2006). The field wind pressure data was applied on the external walls as equivalent mean pressures over an entire face in each direction. Because no pressure data was recorded on the roof the wind pressure applied was an extrapolation of the wind pressure calculated using CFD methods (Appendix E). Again the objective is to assess potential of given types of events to damage transported mini homes. 5.4.5 Numerical modelling results The analysis was executed on a computer with 1GB memory, and it took about 15 minutes to run a static analysis. Figure 37 shows the deformed plot of the home during the factory lifting process. It can be seen that large distortions are predicted in areas located near the window and door openings. During the road transportation process the deformed shape calculated (not shown in this report) was much smaller compared to that during lifting. Tables 8 and 9 provide comparisons of the deformations between the numerical model and field test results. During factory lifting the deformations obtained from the model are larger than test results. However, during the factory yard lifting some sensors recorded larger deformation values than the model predicts. This is thought to be because differential movement between lifting supports during factory-yard lifting was not considered in the model (Model-a). Future model refinement will correct that. It should be noted that some recorded and calculated deformations are very small (less than 0.05 mm) indicating that the movements in the plasterboard are practically zero. During transportation processes deformations recorded were always larger than the model predicted (Model-b). This is because the present model did not incorporate the effects of pre-damage (cracks) already initiated during lifting processes. Inclusion of initial damages in the model during transportation is feasible but would be very challenging considering the multi-scale issues involved. However, as can be seen in Table 8, the current representation of deformations during lifting process is accurate. Overall it can be noted that the outcome from the numerical models are promising enough to answer ‘what-if scenarios’ associated with altering the mini home construction or handling process variables.


Figure 37: Deformed shape of mini home during lifting

Table 8: Comparison of deformations (mm) recorded during field tests and model predictions

(1 = factory lifting; 3 = factory-yard lifting; Model-a=during lifting)

Table 9: Comparison of deformations (mm) recorded during field tests and model predictions (2 = short-transportation; 4 = road test; Model-b=during transportation)

Process A1 A2 A3 A4 A10 A11 B1 B2 B3 B4 B6 1 0.0022 0.0035 0.0042 -0.004 0.003 -0.0007 0.0049 -0.008 0.015 0.065 0.015 3 0.015 -0.015 -0.038 -0.013 -0.14 0.105 -0.011 -0.02 0.015 0.015 0.02

Model-a -0.048 -0.564 0.028 -0.021 0.035 -0.016 0.008 -0.114 0.005 0.003 0.001

Process A1 A2 A3 A4 A10 A11 B1 B2 B3 B4 B6 2 0.12 0.13 -0.20 0.1 0.84 1.43 -0.01 0.04 0.13 0.12 0.01 4 1.0 -0.65 -1.0 -0.04 -0.7 -0.42 -0.014 0.048 -0.15 0.24 0.03

Model-b 0.018 0.021 0.023 0.006 0.002 0.006 0.012 0.005 0.001 0.003 0.015


5.4.6 Stress analysis

Stress analysis was carried out to locate where there is potential for damage in building components during lifting and transportation of mini homes. As was determined by experiments, areas around window and door openings are locations with high deformation potential, i.e. high stress regions. Here a specific example is given of using structural models to predict stresses in such locations. Analysis presented is for an upper corner of a large door opening at the location where deformation sensor A10 was placed (Figures 26 and 28a). Figure 38 shows predicted contours of effective stress for the plasterboard wall lining caused by factoring lifting. It can be seen that high stress concentration develops at the corner of the door, which correlates with where small cracks formed in practice.

Figure 38: Effective stresses (Von Mises stress contours, eσ ) at location of sensor A10 (MPa) In order to assess whether stress levels in external OSB sheathing or internal plasterboard linings have potential to initiate local damage (like cracking) it is necessary to establish a failure criterion against the predicted stresses are compared. Plasterboard is assumed to be a linear isotropic material with the same strength properties in tension and compression. Therefore the well known Von Mises, or Distortional Energy, failure criterion can be used to assess the likelihood of failure (Boresi and Schmidt, 2003). Local failure is predicted when the distortional strain energy in a material reaches the value for failure under


simple tension or simple compression conditions. Von Mises Stress ( eσ ) is determined at any location in OSB or plasterboard using predicted stress components from the relationship:

2222 )13()32()21(2 σσσσσσσ −+−+−=e (11)

Where: σ1, σ2, σ3 = 3-D principal stresses, which are a function of the total state of stresses. The Stress Function (k) is:

k = σe / material strength (12) If k < 1.0 it is predicted that the material is undamaged at the location of interest. If k ≥ 1.0 it is predicted that the material has failed locally. For the purposes of the analysis it was assumed that the strength in tension or compression of plasterboard is 3.14 MPa (Gypsum Association, 2005). OSB is assumed to be a linear elastic orthotropic material and in such a case the Tsai-Wu can be used to assess the likelihood of failure at various locations (Tsai and Wu, 1971). In this criterion the failure envelope is a function of various ratios of applied stresses to material strengths in the principal material directions or planes. In the analysis the 2-D form of the criterion was used based on the assumption that in-plane deformation was the dominant cause of any damage: kFFFFFF =+++++ 2211

26662112

222211 21 σσσσσσσ (13)

where: k =stress function, σ1 = σx, σ2 = σy, σ6 = σxy (predicted stresses);

XX CTF 11

1 += ; YY CT

F 112 += ;

XX CTF 1

11 −= ; YYCT

F 122 −= ;

2661S

F =

TX, TY = tensile strength in x and y directions, CX, CY = compression strength in x and y directions and S = shear strength (Table 10). If k < 1.0 it is predicted that the material is undamaged at the location of interest. If k ≥ 1.0 it is predicted that the material has failed locally.


Table 10: OSB Strength Properties (Chui et el, 2005)

Tension (parallel) 15.87 MPa Compression (parallel) 18.02 MPa Tension (perpendicular) 4.47 MPa Compression (perpendicular) 11.60 MPa Biaxial compression 15.87 MPa Shear 9.70 MPa

The range of Von Mises stresses in Figure 38 is between 1.5 and 4.0 MPa. Table 11 shows the range values of k for an element of the FE mesh near the corner of the door opening. Corresponding stress contours for the elements are shown in Figure 39. It can be seen that failure is predicted to initiate in the plasterboard in element number 15479 near location 1. This is in accordance with damage observed during the field test. At the corresponding location in the OSB sheathing the stress factor (k) calculated as per the Tsai-Wu criterion indicated that failure in the OSB (element 8541) is very unlikely. Damage mitigation strategies should include substituting wall lining materials with higher mechanical properties than are typical for plasterboard.

Table 11: Stress factor k values for elements 15479 and 8541 Material

(Element number) Location within

element

Von Mises (k value)

Corner 1 1.25 Corner 2 0.82 Corner 3 0.35

Plasterboard

(15479) Corner 4 0.49

Tsai-Wu (k value) Corner 1 0.019 Corner 2 0.0054 Corner 3 0.0030

OSB (8541)

Corner 4 0.0072


(a) Stress contour for plasterboard (b) Stress contour for OSB sheathing (element 15479) (element 8541)

Figure 39: Von Mises stresses near corner of door opening

The stress analyses were also performed for road transportation, with model input being maximum accelerations recorded as a result of road roughness. In general it was found that the stresses developed during transportation are smaller than those during lifting of units. For example, stress function values calculated at the location of sensor A10 were between 0.2 and 0.3 for plasterboard. Small stress function values (0.05 to 0.10) were also predicted from wind force analyses. 5.4.7 Sensitivity analysis Two major sensitivity studies were conducted in this project. The first was to find optimum support configurations during lifting and transportation. This included: finding optimum arrangements of structural supports (number and location). The second was to find optimum material and building assembly configurations for a prefab building. Both studies were based on the type of large prefabricated mini home used in field tests. Results are therefore not general but they do illustrate the benefits of analytical investigations. Reinforcing techniques already applied by the collaborating company were investigated for their effectiveness with respect to reducing potential for damages. Part of this was to investigate the effectiveness of using a double rim header in the floor framing versus using a single header in the floor framing (Figure 40). Also, the effects of differential movements between lifting support points was simulated to indicate the likely magnitude of irregular supports on non-synchronised lifting jacks. Table 12 summarizes seven different support scenarios with Case 1 being the reference (current practice). For comparative performance, Von Mises stresses in plasterboard were again calculated for the location of sensor A10 (see Section 5.4.6). Complete Von Mises stresses for all the scenarios can

1 2

3 4

1 2

3 4


be seen in Appendix I. Table 13 shows an example of Von Mises stresses between Case 1 (using 3 lifting supports without differential movement) and Case 3 (using 4 lifting supports without differential movements). It can be seen that using four lifting supports along each long wall of the mini home reduces peak stresses 4.31 MPa to 2.01 MPa, which is a 53% reduction. As results in Appendix E show, a simulation with differential movements of supports (Case 2) generates a tremendous amount of stress concentration at the top of the door opening (and other wall openings). The differential jack movements of ±5 mm were estimated to produce stresses of up to 27.4 MPa at the corner of the door opening for lifting arrangement 2. Interestingly, if there were four supports with irregular movements (Case 4) predicted peak stresses were reduced to 4.3 MPa. In reality plasterboard cannot resist any stresses in excess of about 3 MPa and any prediction of higher stresses equates to the simple conclusion that “cracking is to be expected”. Field observations suggest that assuming differential jack movements of ±5 mm is quite reasonable.

Figure 40: Double rim and single rim headers for floor framing

Also, if there are four lifting supports using only a single (nominally 2x12) rim header in the floor framing is acceptable. Using only three lifting supports with and a single rim header (Case 7, Appendix E) generates large stresses compared to using a double rim header (Case 1). But using four lifting supports with single rim header (Case 4) the stress level is reduced to a safer level (1.2 MPa). Overall it can be concluded that using four lifting supports will mitigate damage and save materials, and lessens sensitivity to unevenness in jack movements. For road transportation sensitivity analysis was based on the situation where there is no reinforcement in the rim header of the floor. Scenarios investigated were:

o Assuming earthquake data available in the SAP2000 FE program (library function based on the Uniform Building Code provisions response spectrum.

o Using maximum accelerations collected during road testing, via the ‘time history analysis’ option in SAP2000.

o Using minimum acceleration data collected during road testing, via the ‘time history analysis’ option in SAP2000.

It was found that dynamic stresses expected to be generated during transportation are small compared to stresses created by lifting processes.

Floor joists (2x12)

Single rim header

Floor joists (2x12)

Single rim header

Floor joists (2x12)

Double rim header

Floor joists (2x12)

Double rim header


Table 12: Scenarios for sensitivity study on lifting arrangements Scenario Arrangement Description

01

Three lifting positions and double rim header as per the regular practice in the industry (reference case). A = 10 feet (3.05 m)

B = 27 feet (8.24 m) 02

Three lifting positions with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 27 feet

03

Four lifting positions A = 10 feet B = 18 feet (5.49 m)

04

Four lifting positions with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 18 feet

05

Four lifting positions with only a single rim header in floor frame. A = 10 feet B = 18 feet

06 Four lifting positions with a single rim header in floor frame, and with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 18 feet

A B B A

B B AA

A B BB A

-5mm+5mm -5mm

A AB B B +5mm -5mm +5mm -5mm

A AB B B

A ABB B


07 Three Lifting positions as per the regular industry practice but with a single rim header in floor frame. A = 10 feet B = 27 feet

Table 13: Comparison of stresses in plasterboard for lifting scenarios Cases 1 and 3

Regular with 3 supports Regular with 4 supports Corner 1 4.31 MPa Corner 1

2.01 MPa

Corner 2 2.83 MPa Corner 2 1.47 MPa

Corner 3

1.19 MPa Corner 3

0.68 MPa

Corner 4 1.72 MPa Corner 4 1.29 MPa

Note: Corner locations and numbering refer to element 15479 (Figure 39).

5.5 Summary of findings for handling and transportation From field tests, the key findings are:

• For the tested mini home damages occurred mostly in the walls or ceiling of the open living area, because that section of the building experiences relatively high forces and has lowest rigidity. In general it can be expected that portions of mini homes with an open interior and/or large wall openings will be susceptible to handling and road transportation damage.

• Visible damages were mainly cracks in plasterboard linings of walls or ceilings. The largest deformations, and therefore highest stresses and most initiation of new damages, occurred during lifting operations in the factory yard. This resulted from poor control of the evenness of the ground in the temporary holding area, lack of synchrony in jack movements, and use of insufficient jacking points along the length of the building.

• The delays in switching lifting jacks created twisting moments about the longitudinal axis of the mini home. Those twisting moments cause cracking in ceilings.

• Damages developed during lifting processes can be accentuated during transportation by vibration induced cyclic fatigue processes.

Based on numerical simulations of the structural behaviour it can be concluded that:

• Plasterboard carries more forces than OSB sheathing in proportion to their strengths during lifting processes.

B B AA


• Results shown that significant reductions in stresses and therefore damage can be expected if mini homes are lifted at four jacking points along the long walls, rather than at three points as at present.

• Material economies are possible, and especially in terms of reducing the amount and dimensions of lumber framing components, through application of advanced structural analysis of the type reported here. Those models can optimise decisions across the effects of both lifting and road transportation processes.

UNB5 High Performance Modular Wood Construction Systems Page 81 of 151

6. Conclusions, recommendations and impacts 6.1 Conclusions A. Primary conclusions based on field and numerical study of air leakage and moisture transport in walls of a typical low-rise building having prefab light-frame wall panels are: A1. In an overall sense the wall, that was insulated with a combination of fibreglass matt

inserted between the studs and an external layer of rigid foamed plastic beneath wood clapboard siding, functioned adequately. Heat was easily retained within the building and there was drying of wood wall components after construction, but no excessive accumulations of moisture within the wall components or cavities. However, the butt joint construction interfaces that are located between adjacent wall panels were excessively leaky, and an unnecessary source of energy loss. Design of building systems must focus on both construction joints and wall panels themselves. The major strength of the current study was that it considered both.

A2. From through-the-thickness differential pressure observations on wall panels it is concluded that air leakage occurs mostly in wall panel-to-wall panel construction joints and not by leakage though the wall panels themselves. Therefore any analysis of thermal efficiency that simply considers the intra-panel wall construction will not lead to an accurate understanding of a buildings thermal efficiency.

A3. From the relative humidity recordings there was little evidence of moisture accumulations at wall panel-to-wall panel joints. The dryness of such joints reflected that there are substantial air movements in such joints, which makes them energy inefficient. This conclusion applies to the climate of Fredericton, NB (inland Maritime). Extrapolation to other Canadian or USA climate conditions would be unreliable without further study.

A4. There is need to develop improved construction details for assembling prefab light-frame wall panels, with emphasis on reducing air leakage and without causing moisture accumulations within walls. In essence, although a healthy indoor climate requires provision of adequate ventilation for indoor spaces, that ventilation should not occur in an uncontrolled manner through the construction joints. Solutions of wall performance problems should not be dealt with separately from design of appropriate HVAC systems.

A5. The field monitoring has resulted in collection of a unique data set for assessment of the hygrothermal performance of a typical wall system and for development and verification of numerical hygrothermal models. That data resource is available via this report.

A6. The two-dimensional numerical hygrothermal model created, based on finite element analysis, is an accurate predictor of heat flows and moisture transport within light-frame wall panels and their construction joints. Applying the model it is possible to predict long term heating efficiencies and moisture performances of alternative construction details, and thereby to optimise solutions.

A7. What optimal hygrothermal design solutions are will depend on the intended service situation and decisions should reflect this. Optimal construction details for light-frame walls will vary depending on both indoor and outdoor climates and seasonal variations in either. Optimization designs are those that make proper consideration of what heat flows, airflows and moisture balances at surfaces will be at the construction location.

B. Primary conclusions based on field and numerical study of typical handling and road transportation practices for prefabricated buildings, such as mini homes, are:


B1. For the typical mini home that was investigated in detail it is not uncommon for damages like cracking of plasterboard to occur between factory assembly and site installation. Although the damages observed would not significantly impair the short term structural performance, they could adversely impact long term structural and non-structural performance, and cause unfavourable customer perceptions of the quality of prefab construction in Canada.

B2. The dominant cause of damages in the studied mini home was inadequate lifting practices during post assembly-line handling at the factory. Road transportation could propagate that damage but only because it was pre-existing.

B3. Based on application of numerical structural analysis models it is possible to design improved handling and road transportation strategies, and to optimise the structural design, including achieving material efficiencies.

B4. Numerical structural analysis models developed during the project yield reasonable results, and can be used to identify better structural and construction practices applicable during factory handling, road transportation, on-site handling and service after completion.

6.2 Recommendations for future work A. In the building physic aspect, recommended future work is: A1. Accelerated hygorthermal tests under laboratory conditions to verify numerical prediction

concerning optimized designs of wall panel-to-wall panel (and other) construction joints for prefab buildings.

A2. Extended development of the hygrothermal model (for heat flow and moisture transport), and extended verification of it against laboratory tests.

A3. Field monitoring of buildings that covers a broader range of outside climatic conditions (e.g. far north, prairies), alternative indoor climatic conditions (e.g. occupied residential and industrial buildings), and altered constructions (e.g. barrier free constructions – without non-storage layers).

B. In the handling and transportation aspect, recommended future work is: B1. Laboratory experiments (wind tunnel study, shaking/vibration tests) to simulate handling

and road transportation processes for prefab buildings. Using controlled conditions would facilitate unambiguous definition of the physical processes.

B2. Development of active or passive vibration absorbers (dampers) that can reduce dynamic forces generated during factory or site handling and road transportation. Such dampers need not be permanent fixtures of the prefab buildings.

B3. Extended sensitivity studies aimed at optimizing construction details and working practices that mitigate possibilities of pre-service damage to prefab wood constructions like mini homes and building modules.

Note: Recommendations have already been made to the collaborating prefab company for methods that will avoid damages to the largest type of mini homes they manufacture. These mainly concern avoiding poor support and jacking methods during handling of mini homes in the factory yard and on site. They have taken appropriate steps to remedy the situation.


6.3 Impact/benefits to the prefab wood industry Impacts/benefits to the prefab wood industry resulting from findings of this project are: • Creation of a technical database of long-term performance of a ‘typical’ prefab wood

building construction. • Creation of advanced numerical hygrothermal models supporting development of innovative

panellized wood buildings. • Creation of a technical database for the structural performance of a ‘typical’ prefab mini

home during handling and road transportation. • Creation of advanced numerical models to facilitate development of innovative framing

systems and of handling and transportation technology for the prefab wood buildings. • Assistance to corporations in Canada. The products of this work have already been, and will

continue to be, offered to Canadian industry by staff from the University of New Brunswick on a fee for services basis.


7. References ASHRAE - American Society of Heating, Refrigerating, and Air-Conditioning Engineers. (1997). Handbook 1997 Fundamentals, ASHRAE, New York, NY. Asiz, A., Iranpour, M. and Smith, I. (2005) Analysis Of Structural Stresses During Handling And Transportation Of Factory-built Housing Construction. Proceedings of 33rd Annual Conference of the Canadian Society for Civil Engineers, Vol. 2, CD-ROM, 8 pp. Boresi, A.P., Schmidt, R.J. (2003), Advanced Mechanics of Materials, 6th Edition, John Wiley & Sons, Inc. Burch, D.M. and Thomas, W.C. (1991) An Analysis of Moisture Accumulation In A Wood Frame Wall Subjected To Winter Climate, Final Report, the U.S. National Institute of Standard and Technology, Gaithersburg, MD. CSA - Canadian Standard Association. (1992) Manufactured Housing/Mobile Homes, Standard CAN/CSA Z240, CSA, Ottawa, ON. CSI - Computers and Structures, Inc. (2005) SAP2000 Nonlinear version 10.1.1 – Analysis Reference Manual, CSI, Berkeley, CA. Cramer, S. M., Friday, O.M, White, R.H. and Sriprutkiat, G. (2003) Mechanical Properties Of Gypsum Boards At Elevated Temperatures, Technical Report, USDA Forest Service Forest product laboratory, Madison, WI. Creighton, J. (1997) Finite Element Analysis Of Manufactured Homes Under Lateral Loading, MS Thesis, Dept of Civil Engineering, Colorado State University, Ft. Collins, CO. Gypsum Association. (2005) Gypsum Board Typical Mechanical and Physical Properties (GA-235-05) www.gypsum.org Johnson, R. J. (1982) Residential Moisture Conditions-Facts And Experience, Moisture Migration In Buildings, pp. 234-240, ASTM STP 779. M. Lieff and H. R. Trechsel, Eds. American Society for Testing and Materials, Philadelphia, PA. Kipp, W.I. (2000) Vibration Testing Equivalence, Proceeding International Safe Transit Association, 26 April 2000, Orlando, FL. Kumaran, M. K. (2006) A Thermal And Moisture Property Database For Common Building And Insulation Materials, ASHRAE Transactions, 112 (2), pp 485-497. Lstiburek, J. and Carmody J. (1996) Moisture Control Handbook, Principles And Practices For Residential And Small Commercial Buildings, Wiley & Sons, New York, NY.


Marcondes, J. and Singh, P. (1992) Use Of Road Roughness To Predict Vertical Acceleration In Truck Shipments, Advances in Electronic Packing, American Society of Mechanical Engineers, pp 999-1004. Mohammad, M. (2006) Connection Systems For Prefabricated Wall Panels, Final Report, No.FCC25, Natural Resources Canada, Ottawa, ON. Chui, Y.H., Pirzada, G., Lai, S. (2005) Enhancing Shearing and Bearing of Wood I-Joist, Final Report, No. UNB3, Natural Resources Canada, Ottawa, ON. Mi, H. (2005) Behaviour Of Unblocked Shear Wall, Master of Science Thesis, Graduate Academic Unit of Forestry and Environmental Management, University of New Brunswick, Fredericton, NB. NAHB – National Association of Home Builders Research Center. (2002) Advanced Panellized Construction - Year One Progress Report, prepared for Partnership for Advancing Technology in Housing (PATH), Washington D.C., U.S. PATH Inventory. (2003) Modular Multiple Dwellings, Partnership for Advanced Technology in Housing, Washington, D.C. Peleg, K. (1984) Impact And Vibration Testing of Shipping Containers, Journal of Sound and Vibration, Vol. 93, Issue No. 3, pp. 371-388. Porter, W.A. (2003) Moisture Transport In Attic Spaces Located in Hod-Humid Climates, Doctoral Dissertation, Agricultural and Biological Engineering Department, University of Florida, Gainesville, FL. Richards, R.F., Burch, D.M. and Thomas, W.C. (1992) Water Vapor Sorption Measurements Of Common Building Materials, ASHRAE Transactions, 98(2), pp 475-485. Rose, W.B. and Francisco, P.W. (2004) Field Evaluation of the Moisture Balance Technique to Characterize Indoor Wetness, ASHRAE Report, Illinois, 28 pp. Schmidt, R. J., Goodman, J. R., Richins, W. D., Pandey, A. K. and Larson, T. K. (2000) Improved Design Of Manufactured Homes For Hazardous Winds, Proceedings of World Conference on Timber Engineering, 31 July - 3 Aug, Whistler Resort, BC. Singh, P., Jarimpas, B. and Saengnil, W. (2006) Measurement And Analysis Of Vibration Levels In Commercial Truck Shipments In Thailand And Its Impact On Packaged Produce, Journal of Testing and Evaluation, Vol.34 No.2. Smith, I., Asiz, A., Dick, K., Doudak, G. and Mohammad, M. (2006) Improving Design Concepts And Methods Through Field-Monitoring Of Timber Buildings, World Conference on Timber Engineering, Aug. 7-10, Portland, OR, CD-ROM, 8 pp.


Splittgerber, H. (1978) Effect Of Vibration On Building And Occupants Of Buildings, Conference on Instrumentation for Ground Vibration and Earthquakes, 147-152, Institution of Civil Engineers, London, UK. Straube, J., Onysko, D. and Schumacher, C. (2002) Methodology And Design Of Field Experiments For Monitoring The Hygrothermal Performance Of Wood Frame Enclosures, Journal of Thermal Environments and Building Science, Vol. 26, N2. Taylor, J.I. (1994) The Vibration Analysis Handbook: A Practical Guide For Solving Rotating Machinery Problems, 1st ed., Vibration Consultants, Inc, Tampa, FL.33611 Taynor, T. (2007) AB Exclusive State of the Industry Report for 2006: Total Housing Up 3/7% to 3.024 Millions Units, Automated Builder Magazine, January Edition, pp 8-10. Tenwolde, A., Carl, C., and Malinauskas, V. (1995) Airflow Performance Of Building Envelopes, Components, And Systems, ASTM STP 1255, Mark. P Modera and Andrew K. Persily, Ed., American Society for Testing and Materials, Philadelphia, PA, pp.137-155. Tsai, S.W. and Wu, E. M. (1971) A General Theory Of Strength For Anisotropic Materials, Journal of Composite Materials, 1: 58-80. Winkel, M.H. (2006) Behavior Of Light-Frame Walls Subject To Combined In-Plane And Out Of Plane Loads, Master of Science Thesis, Graduate Academic Unit of Forestry and Environmental Management, University of New Brunswick, Fredericton, NB.


Appendix A Temperatures


Sensors numbering system

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

10

11

12

10

11

12

16

17

18

16

17

18

1

2

3

1

2

3

4

5

6

4

5

6

7

8

9

7

8

9

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m


Note:

• Horizontal axis labeling (month): o 1=May, 2=June, 3=July, 4=Aug, 5=Sept, 6=Oct, 7=Nov, 8=Dec, 9=Jan,

10=Feb, 11=Mar, 12=April.

Period April 2005- April 2006


Period April 2006- Feb 2007


Appendix B Relative Humidity

• Channels: 19 and 20 (joint) • Channels: 36 and 39 (wall cavity)



10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

10

11

12

10

11

12

16

17

18

16

17

18

1

2

3

1

2

3

4

5

6

4

5

6

7

8

9

7

8

9

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m


Appendix C Differential Pressures



10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

10

11

12

16

17

18

1

2

3

4

5

6

7

8

9

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

10

11

12

10

11

12

16

17

18

16

17

18

1

2

3

1

2

3

4

5

6

4

5

6

7

8

9

7

8

9

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

13

14

15

24 25

21 23

22 26

19

2039

36

Butt-joint Cavity

studs

top plate

bottom plate

2.44 m

2.40

m


Appendix D Finite Element Formulation


Finite Element Discretization

Formulation: By integrating the governing Equations (1) and (2) described in the main report over the whole domain (V), and introducing the basis functions in each integral term, results in the weak form equations:

∫∫⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

∂∂

⎥⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡∂∂

⎥⎦

⎤⎢⎣

⎡V

x

TxMx

i

i

Vpwoodi

i

xTk

xTD

xMD

xNN

TCM

tNN

00

00

ρ

∫⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

∂∂

⎥⎦

⎤⎢⎣

⎡+

Vy

TyMy

i

i

yTk

yTD

yMD

yNN0

0 (D1)

Following the procedure of integration by parts, Equation (D1) can be expressed in term of volume and surface integrals:

dy

yTk

yTD

yMD

NN

y

dS

yTk

yTD

yMD

NN

dx

xTk

xTD

xMD

NN

x

dS

xTk

xTD

xMD

NN

TCpwoodM

tNN

Vy

TyMy

i

i

Sy

TyMy

i

i

Vx

TxMx

i

i

Sx

TxMx

i

i

Vi

i

∫

∫∫

∫∫

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

⋅⎥⎦

⎤⎢⎣

⎡∂∂

−

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

⎥⎦

⎤⎢⎣

⎡+

⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

⋅⎥⎦

⎤⎢⎣

⎡∂∂

−⎟⎟⎟⎟

⎠

⎞

⎜⎜⎜⎜

⎝

⎛

∂∂

∂∂

+∂

∂

⎥⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡∂∂

⎥⎦

⎤⎢⎣

⎡

00

00

00

00

00

ρ

(D2)

where i = number of nodal points for an element, Ni = weighting (shape) functions, and S = boundary surface. Utilizing boundary conditions stated in Equations (3) to (6) of the main report, and using the


approximate functions for time dependent moisture and temperature as given in Equations (D3) and (D4), then substituting these finite element approximations into the weak form Equation (D2), results in the finite element discretization Equation (D5).

( ) [ ]{ }MNyxM =, (D3)

( ) [ ]{ }TNyxT =, (D4)

[ ] [ ] [ ]∑ ∫∑ ∫∑ ∫ =⎭⎬⎫

⎩⎨⎧

+⎭⎬⎫

⎩⎨⎧

∂∂ dSB

TM

dVKTM

tdVC

See

e

Vee

e

Ve (D5)

where;

[ ] ⎥⎦

⎤⎢⎣

⎡=

jip

ji

NNCNN

Cρ0

0 (D6)

[ ] ⎥⎦

⎤⎢⎣

⎡+

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡

∂∂

∂∂

∂∂

∂∂

∂∂

∂∂

+

⎥⎥⎥⎥

⎦

⎤

⎢⎢⎢⎢

⎣

⎡

∂∂

∂∂

∂∂

∂∂

∂∂

∂∂

=ji

ji

jiy

jTy

ijMy

i

jix

jTx

ijMx

i

hNNSNN

yN

yNk

yN

Dy

Ny

ND

yN

xN

xNk

xN

Dx

Nx

ND

xN

K0

0

0

(D7)

[ ] ⎥⎦

⎤⎢⎣

⎡=

∞

∞

hTNSMN

Bi

i (D8)

where: i, j = 1 - 4, if four node elements are used.

Solving time-dependent equations: To solve for time dependent problem in Equation (D5) a forward time difference is applied. Rearranging Equation (D5) yields:

[ ] [ ] [ ] [ ]n

e

enn

e

e

TM

KCt

BTM

Ct ⎭

⎬⎫

⎩⎨⎧

⎟⎠⎞

⎜⎝⎛ −

∆+=

⎭⎬⎫

⎩⎨⎧

⎟⎠⎞

⎜⎝⎛

∆+

+11

21

1

(D9)

where: ∆t = time step, and n = increment number. A computer code was developed to solve the above time dependent algebraic equations. In the


model a lumped capacitance matrix was used for [C]. This method diagonalizes the matrix by replacing each diagonal with the sum of the terms on its row. Thus,

( ) ∫∫ =∑=ee V epiV ejpi

NE

jdVCNdVNCNiiC ρρ, (D10)

( ) jijiC ≠= ,0, (D11) If

Replacing [C] with a diagonal matrix means that the capacity of the moisture and temperature parameters (M and T) would be concentrated at the nodes rather than distributed throughout the elements. Diagonalization, therefore, means that a distributed capacitance is replaced by many point capacitances. Stable time increment: When lumped capacitance is used to solve unsteady state equations the algorithm can become unstable. The following equations are used to check whether the time step (∆t) or meshing size (∆x) taken will result in stable solutions:

Rxt

2)( 2∆

<∆β , producing stable solution with no oscillations.

,2

)( 2

Rxt ∆

>∆β producing stable solution, but oscillation may occur.

R

xt2)(∆

>∆β , unstable, oscillation with increasing magnitude may occur.

where R = either the diffusivity coefficients in the moisture equation or the thermal conductivities in the energy equation, β = heat capacity in the energy equation, ∆x = diameter of a circle circumscribing the element at the point being considered.


The following chart is summary of the main computer coding for solving heat and moisture transfer in a building assembly.

input from pre-processors: geometry (MESH) and B.C (indoor and outdoor climate conditions)

for each increment of time

for all elements in the domain (MESH)

for all quadrature points in an element

Create element contributions to [C], [K], {B}

Place current M,T values in {M, T} array

if values should be saved

Solve [C]{M, T(t+dt)} = {B} + [K]{M, T(t)} for {M, T(t+dt)}

Save {M, T} array and go to post-processing (plotting)

define initial conditions and time step

call the basis (shape) function


Appendix E Computational Fluid Dynamics (CFD) Analysis

of Prefab Mini Home During Transportation


Introduction The main purpose of this activity is analytical study of wind pressure patterns on a prefab mini home while it is being transported, prior to actual wind pressure measurements to assist the design of field tests. There are always some difficulties associated with field measurements of wind pressures, such as limitations on the number of sensors available and finding the best spots on the building surfaces to mount pressure taps. Numerical approaches can be employed to investigate fluid (air) flow behaviour around objects like buildings. In this study a Computational Fluid Dynamics (CFD) scheme is used to model the fluid flow around a moving tractor-trailer carrying a mini home. A 3-D model is prepared and the CFD simulations are carried out for different truck velocities and the pressure distributions on the walls determined. This permitted deductions to be made about the best positions of pressure sensors / taps in real road tests. Important details of the CFD model and results are given here.

Geometrical Model Geometrical modelling is done for individual components of the system. Components are the tractor trailer (truck), mini home, and a relatively large wind tunnel-like domain through which air flows past the other two components that are positioned in juxtaposition within that larger domain. After finishing the geometrical modelling for each component, they are assembled together in a single Computer Aided Design (CAD) file that is exported to mesh generator software. A schematic of the truck and mini home is depicted in Figure 1.

Figure 1: 3D model of the truck and prefab house

Mesh Generation As the geometry of the case study is rather complex, unstructured tetrahedral volume finite elements were chosen to generate an appropriate mesh for the CFD simulation. Local mesh size refinement was taken into consideration to decrease the numerical inaccuracies. Figure 2 shows the surface mesh generated for the truck and mini home (house).


Figure 2: Surface mesh generated for truck and house

The total number of volume elements for the whole domain is approximately 435,000. The model contains a significant volume of air around the truck and the house to account for boundary layer effects on air flow, as in physical boundary layer wind tunnel tests. Boundary Conditions Figure 3 shows the schematic boundary conditions considered in the analysis. As noted, the velocity boundary condition is considered for the inlet section of the domain. Velocities considered were 90 and 110 km/hr, which are close to actual maximum speed conditions of trucks on roads but presumes still air conditions. So far only symmetric boundary conditions have been considered for the sides, top and outflow boundaries.

Figure 3: Schematic of the boundary conditions applied

inlet

outletTransported prefab house

Boundary layer


Mathematical Model The finite volume method was used to discretize the continuity and momentum equations governing the fluid flow. The turbulent nature of the flow is treated by the standard k-ε model. The k-ε model is the most commonly used of all the turbulence models. It is classified as a two- equation model. This denotes the fact that the transport equation is solved for two turbulent quantities k and ε. One for the turbulent kinetic energy k and a further one for the rate of dissipation of turbulent kinetic energy ε, are solved. Setting up equations of continuity, x-momentum, y-momentum, z-momentum, k and ε, their concurrent solution of them produces the distribution of the pressure, fluid velocities in x, y, z directions, k and ε that can be post-processed to extract the variables of interests (Malalasekera et al 1995; Chung 2002). The convergence of the numerical scheme is ensured by tracking the residual values of the solved variables and physical quantities such as velocity at the outlet section and the skin friction coefficient. The boundary layer produced drag is often referred to as skin friction. After ensuring computational accuracy of the numerical scheme and convergence of the quantities, the results have been prepared for post-processing. The model is implemented via the well known commercial FLUENT CFD software (Fluent, 2005). Results As the objective of this work was to investigate pressure distributions on the mini home walls, emphasis was placed on those regions. It is predicted that the front wall of the house is the most susceptible region for wind pressure, which is expected as it is directly facing oncoming wind motion. Figure 4 shows the predicted pressure distribution on the front face of the mini home. As seen in the picture, there are substantial pressure gradients. In general the pressure is highest toward the bottom edges of the wall and decreases toward the centre which is behind the cab of the truck.

Figure 4: Contour of pressure on the front face of the house


The region of maximum pressure is indicated with letter A in Figure 4. This region has the static pressure of 645 Pa (at 110 km/hr). The minimum pressure on the same face was about -440 Pa (at 110 km/hr). These values indicated the likely range of pressures to be measured experimentally and also pointed out probable locations of highest pressure gradient. Similar analysis was done for other faces of the house to find the critical pressure regions. The contours of the static pressure on the back face, side faces and roof faces are depicted in Figures 5, 6 and 7 respectively.

Figure 5: Contour of pressure on the back of the house Figure 5 shows that there is a high-pressure region at the centre of the back face of the mini home. However, the magnitude of the pressure gradients is predicted to be less than for the front face. This is due to the wake region which appears behind the truck and house. There is asymmetry in the pressure contours in that region which was an artefact of numerical inaccuracies. Although this needs to be investigated further, the general trends are reasonable. Note: In practical situations there will be considerable turbulence around the mini home because neither the objects (truck and house) nor the wind flow are ever truly symmetrical. Plotting the pressure contours at the side walls faces of the house indicates that the pressure gradients at those regions do not have drastic variations and pressures change quite smoothly. Figure 6 shows this. Pressures on the roof of the house are predicted to be almost the same as on side walls. It is important however to realise that any conclusions here only apply to a truck driving in still air or parallel with the environmental wind.


Figure 6: Contour of pressure on the side of the house

Figure 7: Contour of pressure on the roof of the house

References: Chung, T.J (2002). Computational Fluid Dynamics Cambridge University Press, Trumphington street, Cambridge ,UK. Fluent, Inc. (2005), Computational fluid dynamic FLUENT version 6.2, Lebanon, New Hampshire. H. K. Versteeg and W. Malalasekera (1995). An introduction to computational fluid dynamics - The finite volume method, Longman Scientific & Technical, Longman Group Limited , Longman House, Burnt Mill, Harlow Essex CM20 2JE, England.


Appendix F Selected Deformation Results During Lifting and Handling

Notes:

xx-01: Lifting at the factory xx-02: Short transportation between the factory and yard xx-03: Lifting (re-loading) to the actual truck xx-04: Road test xx: Numbering system for the deformation sensors (example: A11-01 shows

deformation at sensor A11 during lifting at the factory)


Appendix G Wind Speed, Direction and Pressure


Wind pressure range recorded, for all differential sensors


Appendix H Acceleration/Vibration Results


Table 1: Statistical analysis of accelerations

Locations Lowering in factory yard

Lifting in factory yard

Local roads Highway

Min (g) -0.02 -0.05 -0.34 -0.45

Max (g) 0.03 0.06 0.36 0.47 Mean (g) 0.00 0.01 0.00 0.00

C0

Std Dev. 0.01 0.01 0.07 0.11 Min (g) -0.01 -0.03 -0.33 -0.41

Max (g) 0.06 0.08 0.36 0.50 Mean (g) 0.03 0.02 0.03 0.05

C2

Std Dev. 0.01 0.01 0.05 0.06 Min (g) -0.11 -0.12 -0.75 -0.68

Max (g) -0.04 0.06 0.49 0.45 Mean (g) -0.08 -0.04 -0.12 -0.12

C3

Std Dev. 0.01 0.01 0.06 0.09 Min (g) -0.06 -0.05 -0.41 -0.55

Max (g) -0.01 0.02 0.38 0.47 Mean (g) -0.04 -0.02 -0.04 -0.06

C1

Std Dev. 0.01 0.01 0.05 0.08


-0.40

-0.35

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.001 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 1: Plot of the minimum acceleration during lifting in the factory yard

-0.10

0.00

0.10

0.20

0.30

0.40

0.50

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 2: Plot of the maximum acceleration during lifting in the factory yard


-0.05

-0.04

-0.03

-0.02

-0.01

0.00

0.01

0.02

0.03

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Elapsed time (16s)

Acce

lera

tion

(g)

Ch00Ch02Ch03Ch11

Fig. 3: Plot of the mean acceleration during lifting in the factory yard

0.000.010.010.020.020.030.030.040.040.05

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 4: Plot of the absolute mean acceleration during lifting in the factory yard


-1.40

-1.20

-1.00

-0.80

-0.60

-0.40

-0.20

0.001 11 21 31 41 51 61 71 81 91

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 5: Plot of the minimum acceleration on local roads

0.000.020.040.060.080.100.120.140.160.18

1 11 21 31 41 51 61 71 81 91

Elapsed time (16s)

Acc

eler

atio

n (g

) Ch00Ch02Ch03Ch11

Fig. 6: Plot of the absolute mean acceleration on local roads


-0.14-0.12-0.10-0.08-0.06-0.04-0.020.000.020.040.06

1 11 21 31 41 51 61 71 81 91

Elapsed time (16s)

Acc

eler

atio

n (g

)Ch00Ch02Ch03Ch11

Fig. 7: Plot of the mean acceleration on local roads


0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1 11 21 31 41 51 61 71 81 91 101

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 8: Plot of the maximum acceleration on highway

-1.20

-1.00

-0.80

-0.60

-0.40

-0.20

0.001 11 21 31 41 51 61 71 81 91 101

Elapsed time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 9: Plot of the minimum acceleration on highway


0.000.020.040.060.080.100.120.140.160.18

1 11 21 31 41 51 61 71 81 91 101

Elasped time (16s)

Acc

eler

atio

n (g

)

Ch00Ch02Ch03Ch11

Fig. 10: Plot of the absolute mean acceleration on highway


0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

Elapsed time (16s)

RMS

(g2)

Ch00Ch02Ch03Ch11

Fig. 11: Plot of the root-mean-square acceleration in 1/3 octave band around the first resonant frequency – lowering at factory yard

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

0.080

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Elapsed time (16s)

RM

S(g2

)

Ch00Ch02Ch03Ch11

Fig. 12: Plot of the root-mean-square acceleration in 1/3 octave band around the first resonant frequency – lifting at factory yard


0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

0.450

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

Elapsed time (16s)

RM

S (g

2)Ch00Ch02Ch03Ch11

Fig. 13: Plot of the root-mean-square acceleration in 1/3 octave band around the first resonant frequency – local road

0.00

0.05

0.10

0.15

0.20

0.25

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Elapsed time (16s)

RM

S (g

2)

Ch00 Ch02Ch03 Ch11

Fig. 14: Plot of the root-mean-square acceleration in 1/3 octave band around the first resonant

frequency – highway


Appendix I Stress Analysis for Various Lifting Scenarios


Notations for element being investigated

Scenarios for lifting

Scenario Arrangement Description 01

Three lifting positions and double rim header as per the regular practice in the industry (reference case). A = 10 feet (3.05 m)

B = 27 feet (8.24 m) 02

Three lifting positions with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 27 feet

wOpening (window)

Finite element mesh around opening

1 2

34

Element being investigated, corner labeling

wOpening (window)


wOpening (window)wOpening (window)


1 2

34


1 2

34

1 2

34


A B B A

B B AA

-5mm+5mm -5mm


03

Four lifting positions A = 10 feet B = 18 feet (5.49 m)

04

Four lifting positions with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 18 feet

05

Four lifting positions with only a single rim header in floor frame. A = 10 feet B = 18 feet

06 Four lifting positions with a single rim header in floor frame, and with uneven support movements simulating non-synchronised jack movements. A = 10 feet B = 18 feet

07 Three Lifting positions as per the regular industry practice but with a single rim header in floor frame. A = 10 feet B = 27 feet

B B AA

A B BB A

A AB B B +5mm -5mm +5mm -5mm

A AB B B

A ABB B


Case 1- Regular 3 handling supports at each side

Von Mises stresses around A10 sensor location (GWB)

Corner 1 4.3132 MPa Corner 2 2.8277 MPa Corner 3 1.1850 MPa Corner 4 1.7174 MPa

Case 2- Regular 3 supports with simulated differential ground displacements




Case 3- Regular 4 supports on each side

Von Mises stresses around A10 sensor location (GWB) Corner 1 4.3132 MPa Corner 2 2.8277 MPa Corner 3 1.1850 MPa Corner 4 1.7174 MPa

Case 4- Regular 4 supports with simulated differential ground displacements




Case 5- Regular 4 supports with 1 header rim



Case 6- Regular 4 supports with 1 header rim and differential ground displacements




Case 7- Regular 3 supports with 1 header rim

Von Mises stresses around A10 sensor location (GWB) Corner 1 5.4669 MPa Corner 2 3.2467 MPa Corner 3 1.4714 MPa Corner 4 2.5059 MPa

High Performance Modular Wood Construction Systems · Project No. UNB5 Value to Wood Project...

Documents

Transcript of High Performance Modular Wood Construction Systems · Project No. UNB5 Value to Wood Project...