PONCHO
Pro-active Observational Network for Capillary and Hygroscopic Optimization
In collaboration with Joel Esposito
,
This ongoing research focuses on the physics and phenomenology of moisture intrusion in wood frame construction. The primary objective is to determine the degree to which fiber cement siding direct applied to wood structural panels is problematic, as this assembly design is widespread in Northeast Florida and is currently considered to be compliant with the IRC. The research is being conducted as a part of Joel's thesis, tentatively titled Fiber Cement Cladding Intimately Attached to Wood Structural Panels: Analyzing the Extent of Risk. The research efforts culminate in a proposition for a machine learning system (PONCHO) to detect or predict when risk conditions may occur.
Code Analysis
According to IRC R703.7.3.2, “water-resistive barrier(s) (WRB) shall have a drainage efficiency of not less than 90 percent, as measured in accordance with ASTM E2273…”. ASTM E2273 is a testing standard for moisture drainage that uses Exterior Insulation and Finishing System (EIFS) as a cladding layer. This means that EPS insulation (one layer of the EIFS cladding) is in direct contact with the wood structural panel (WSP). A major manufacturer of WSP’s has successfully tested their product to using ASTM E2273. However, they use mechanical fasteners to attach the panels (not adhesive). Thus, the microstructure texture of an integral WRB can effectively convey water down the face of the panel. However, if the drainage paths facilitated by EPS are less diffuse than those created by other products such as fiber cement, this is not a “worst case scenario” test. EIFS is not the only cladding system in the light frame construction market. Wood frame construction is often clad in direct-applied fiber cement siding (including panel siding). If fiber cement siding absorbs excess moisture due to wind-driven rain, a construction defect, or other mechanism that allows water ingress behind the cladding layer, risk will be introduced to the durability and longevity of the building enclosure. This has the potential to impact quality of life for building occupants and the potential to impart economic hardship on those who can least afford to bear it. It is thus imperative to quantify the risks that this assembly type poses to existing wood frame construction.
Microstructure texture on face of integral WRB.
Structured Review
Before engaging in the project, a structured review was conducted to investigate the physics of moisture migration, material behavior, and facade drainage and drying. The results suggest fiber cement does in fact exhibit slower drying times than non-absorptive materials and thus can act like a “sponge”, holding water against the structural substrate. The implications of this are important as according to the current knowledge of the researcher the EIFS testing is not benchmarked or contextualized against other potentially more absorptive materials, nor those with a potentially greater moisture carrying capacity, such as fiber cement.
Broad Potential Scope
Screenshots from a UF CityLab IPAL visit to a multi-story residential project using conventional building wrap. The high exposure and minimal overhangs indicate elevated risk, especially if absorptive cladding is used without a rain screen.
Drainage vs. Drying
Key takeaway: just because moisture drains quickly at the face of the WRB (per ASTM E2273) does not mean there is no risk. Moisture absorbed by the cladding (sponge-like behavior) can be subsequently distributed into the wall via solar drive and other moisture transport mechanisms (capillary action, hydrostatic pressure, etc.).
The Cost of Failure
Above is a photo of a failed facade in Northeast Florida. The occupants have attempted a patchwork remediation, but appear to be unwilling or unable to bear the cost of recladding and are thus essentially stuck living in a house with potential structural degradation. Although this is an improperly designed and executed stucco-on-frame house, it illustrates the importance of a functioning moisture drainage plane. Moisture sensors placed at high risk areas could have notified the occupants of failure prior to widespread damage occurring and afforded reaction time to develop a stopgap remediation strategy. The result of envelope-wide failure is wealth destruction, wasted materials and labor, unnecessary refuse upon the inevitable demolition of the structure, and negative impact on property values and thus revenue for the municipality.
Economic Utility
A vision for a robust envelope moisture monitoring experiment is proposed. The design of experiment is as follows:
PHASE I - Graduate Thesis Project
A) Moisture sensing devices (Raspberry Pi or similar)
B) Data receiver (smart device, cloud storage, etc.)
C) Post-processor (machine learning)
PHASE II - Marketable Intellectual Property (Post-graduate Research?)
D) Output/alert system (mobile or cloud application)
E) Data aggregation (predictive model)
The post processing of data with machine learning techniques such as feature extraction can elucidate complex relationships between variables.* If deployed on a mass scale, this system could provide valuable insights for building product manufacturers (WSP, WRB, siding, paint), home inspectors, real estate agents, and insurance underwriters. This is particularly salient in light of steadily rising homeowners insurance rates. Any technology that promotes reduced risk can help to control costs both for the homeowner and the insurer. The results of this data collection would also be of sufficient breadth and detail to inform policy making such as model building codes.
*From querying session with ChatGPT defining the benefits of ML in the proposed application:
Machine learning can: 1) Automate the detection of patterns that may be too subtle or complex for human observation. 2) Enable robust predictive models even with sparse or noisy data, reducing the need for large amounts of data while improving the accuracy of forecasts. 3) Uncover hidden interactions and nonlinear dependencies between features, which may be impossible to detect through simple plotting.
Thus, feature extraction and machine learning can not only enhance the depth and quality of insights you can gain from time-series data, but they also enable you to create robust, data-driven systems that can continuously learn and predict, even with relatively less data or complex environmental conditions.
This preliminary research investigates how to quantify risks associated with a common construction assembly in Northeast Florida. A comprehensive outline for further research was proposed including an experiment design to gather and process data about the phenomenon of interest. This project stands to answer an important building science question, improve envelope design, inform policy making, and provide economic benefits to multiple stakeholders.
Gallery
