Oral Sessions

Session 1: 

Associate Dean Brigid Mullany 

Associate Professor Brett Tempest 

Assistant Professor Jaewon Oh 

Postdoctoral Researcher Yunesh Saulick

Session 2:

Associate Prof Olya Keen 

Professor Ron Smelser 

Associate Professor Taghi Mostafavi 

Postdoctoral Researcher Yunesh Saulick 

In today's landscape, informatization and automatization have emerged as prevailing trends within the architecture-engineering construction and facility management (AEC/FM) industries (Zhang et al., 2022). Among these, artificial intelligence (AI) stands out as a potent technology, progressively revealing its diverse capabilities across various sectors (Regona et al., 2022). Despite the existence of the concept of AI for several decades, the level of understanding and awareness of AI among FM professionals remains unclear. Limited research and knowledge exist regarding how well FM professionals grasp AI concepts, deploy its use, comprehend its potential benefits, and understand the challenges associated with its implementation. Research This research aims to delve into the current landscape of AI understanding and awareness among Facility Management (FM) professionals. The study also endeavors to assess the present utilization of AI within FM practices and explore potential future applications. Additionally, it aims to scrutinize the factors influencing the adoption and implementation of AI in FM settings, analyzing comparative differences across diverse geographic regions or facility types. A survey was designed to collect information on facility managers’ awareness and understanding of selected popular AI tools and resources. This instrument was then pilot-tested by the SMEs to ensure its relevance. In November 2023 an invitation to participate in the survey was sent electronically via Qualtrics to more than 2,000 FM professionals. Over 400 individuals responded to the survey, reflecting a 15% response rate. The survey data was analyzed using a combination of statistical techniques including descriptive frequency counts, charts, and crosstabulations of responses for variables of interest. Preliminary Results: Among the respondents surveyed, 28% are not aware of the use of AI in facility management, 35% have a limited understanding of AI and 18% express a high level of understanding of the concept. Interestingly, only 4% reported formal AI-focused training participation, with an equal percentage expressing disinterest due to its perceived irrelevance in their current roles. However, 78% of those without AI training expressed keenness in engaging with such programs if presented with the opportunity. In practical implementation, only 23% affirmed integrating AI-driven technologies into their facility management practices. Overall, most facility managers (70%) expressed strong support and advocacy for the adoption of AI in facility management. The impact of this research serves as a pivotal catalyst for fostering broader systemic changes in workplace awareness of the integration of AI. Firstly, the findings provide data on the level of understanding of AI among FM professionals. Thereby, acting as a clarion call for organizations to reevaluate preconceived notions on their workforce understanding of AI. The research outcomes provide empirical evidence to support the necessity of providing the workforce with training and educational programs focused on AI applications. Beyond the corporate sphere, the study's implications reach educational institutions, encouraging the incorporation of AI-based curricula that prepare future FM professionals for technologically driven workplaces.

In this paper, we propose a decentralized bi-level optimization approach for multi-robot networks to achieve connectivity maintenance while accounting for Gaussian-distributed localization uncertainty. We first propose the Probabilistic Control Barrier Certificates using Control Barrier Functions (CBF) that can enforce distance-based pairwise inter-robot connectivity and safety with satisfying probability under positional noise. Then by integrating these certificates and task-related controllers into our proposed graph-based Least Constraining Tree (LCT) approach, the lower-level task entails finding the minimally constraining subsets of existing connectivity edges to maintain at each time step, which are used to specify the step-wise set of control constraints, for the purpose of preserving global and subgroup connectivity as robots move. With the upper-level task of optimizing control deviation from the nominal task-related controllers, our bi-level optimization framework thus enables the joint optimization of the connectivity constraints to enforce (specified by the selected edges) and the resultant control revisions, rendering the intended minimally revised multi-robot controllers while satisfying safety and connectivity constraints under uncertainty. This bi-level optimization is step-wise to allow for dynamic connectivity maintenance and provides the greatest flexibility for executing robots' original tasks. Moreover, we propose the Uncertainty-Aware Decentralized Least Constraining Tree (Dec-LCT) algorithm that interleaves the constraint specification from the proposed LCT approach and solving the resulting optimization-based control problem in a fully decentralized manner. Theoretical proofs are provided to justify the optimality of our approach. Simulation results demonstrate the effectiveness of our method with up to 80 robots. 

Detecting SARS-CoV-2 in wastewater provides a cost-effective alternative to expensive methods like random group testing and individual clinical tests, potentially identifying asymptomatic cases. Electronegative Membrane Filtration (EMF) is a widely used, cost-effective tool for large-volume wastewater virus concentration. However, there's limited research on efficient RNA extraction methods. This study aimed to achieve efficient RNA extraction kits for wastewater surveillance by integrating them with the EMF method of virus concentration. Raw wastewater samples from three treatment plants were collected in Mecklenburg County. Two RNA extraction protocols, the 'QIAamp Viral RNA Mini Kit' and the 'Zymo Quick RNA Viral Kit,' utilizing lysis buffer principles, were modified to enhance RNA yield by adjusting buffer ratios and by removing inhibitions. These optimized methods were compared with two different extraction methods, the ‘RNeasy Power Water Kits’ and the ‘AllPrep Power Viral DNA/RNA Kit,’ utilizing bead-beating principles. The Zymo Quick RNA Viral Kit outperformed the other three kits significantly in Cq value (29-31), copies per reaction, and copies per liter (100,000 to 350,000, P<0.01). It also exhibited a better recovery of 10% of the surrogate Bovine Coronavirus (BCoV). Moreover, the ‘Zymo kit’ was found to be cost-effective at $159.5 for 50 preps compared to other extraction methods. This suggests that the ‘Zymo Quick RNA Viral Kit’ is more efficient at extracting and purifying viral RNA, enabling accurate virus quantification. It informs public health decisions and supports the development of sensitive diagnostic tests, crucial for early virus detection and control.

Solid-State Circuit Breaker (SSCB) has been identified as a potential game-changing technology for DC distribution. However, fewer SSCB research and applications have been reported in AC due to the direct competition with the conventional electromechanical circuit breaker, which has a proven record as AC circuit protection device. This paper investigates the current protection and soft start functions of a three-phase SSCB in the motor control center (MCC). The proposed SSCB integrates application-specific protection functions of soft starter, magnetic contactor, circuit breaker, and thermal relay. The paper designs and implements a three-phase SSCB with a rated voltage of 380VAC and a rated current of 63A. Three key design considerations, including semiconductor device selection, metal oxide varistor (MOV) and snubber selection, and heat sink selection are introduced. The current fault protection of the SSCB was tested and verified at 200A by a pulse test platform with a microsecond-level fault current clearing time. Finally, the novel soft start strategy was analyzed and realized without zero current detection. Both simulation and experimental results validate the novel soft-starting method with significant reduction of motor inrush current. By combining the conventional contactor, circuit breaker, and soft starter functions into one, SSCB can offer competitive advantages in MCC applications.

Solid-State Circuit Breaker (SSCB) has been identified as a potential game-changing technology for DC distribution. However, fewer SSCB research and applications have been reported in AC due to the direct competition with the conventional electromechanical circuit breaker, which has a proven record as AC circuit protection device. This paper investigates the current protection and soft start functions of a three-phase SSCB in the motor control center (MCC). The proposed SSCB integrates application-specific protection functions of soft starter, magnetic contactor, circuit breaker, and thermal relay. The paper designs and implements a three-phase SSCB with a rated voltage of 380VAC and a rated current of 63A. Three key design considerations, including semiconductor device selection, metal oxide varistor (MOV) and snubber selection, and heat sink selection are introduced. The current fault protection of the SSCB was tested and verified at 200A by a pulse test platform with a microsecond-level fault current clearing time. Finally, the novel soft start strategy was analyzed and realized without zero current detection. Both simulation and experimental results validate the novel soft-starting method with significant reduction of motor inrush current. By combining the conventional contactor, circuit breaker, and soft starter functions into one, SSCB can offer competitive advantages in MCC applications.

Historically, traditional concrete tests such as slump and strength tests have proven to be very poor predictors of long-term concrete properties. Hence, the need to develop novel tests, to provide a greater degree of correlation to the future performance of concrete. Performance Engineered mixtures (PEM) tests have emerged as a promising solution and has already been implemented on highway pavements, revealing promising results. However, the viability of these tests for relatively thicker airfield pavements has yet to be determined, birthing this research initiative. Interviews with experts in the industry were conducted and their feedback was used to inform the project objectives. The main objective is to establish a practical mixture design tool that contractors and project managers can use to produce airfield pavement mixtures with a 3 response to vibration and incorporate this tool into code for ease of adoption at Airfields nationwide. Ongoing investigations have been focused on identifying the most promising mixtures, using the new PEM tests followed by field validation through shadow testing. Finally, Specifications will be developed for implementing the mixture design tool and tests will be recommended for implementation in the Federal Aviation Administration (FAA P-501) and the military Tri-Services UFGS airfield pavement specifications, underscoring the project’s national significance. By providing tools that can accurately predict the performance of concrete, this project will help contractors, project managers, aviation operators, and the wider community to improve the quality, longevity, and sustainability of millions of acres of airfield pavement. The use of PEM for airfield pavements has the potential to revolutionize the aviation industry as it provides a platform to determine concrete properties before it is even poured on our airfields.

Previous benchtop studies demonstrated infrared (IR) laser sealing and cutting of blood vessels, in a sequential, two step approach. This study describes a smaller, laparoscopic device design, and simultaneous approach to sealing and bisection of vessels. A 1470-nm IR laser sealed and bisected 40 porcine renal arteries, ex vivo. A reciprocating, sidefiring, optical fiber, housed in a transparent square quartz optical chamber (2.7 x 2.7 x 25 mm OD), delivered laser energy over an 11 mm scan length, with a wide range of powers (41-59 W) and treatment times (5-21 s). Vessel diameters ranged from 2.5-4.8 mm. All vessel cut ends were successfully sealed (80/80), as indicated by burst pressures greater than 360 mmHg. The highest power, 59 W, resulted in short times of 5-6 s. Peak temperatures on the external chamber surface reached 103 degrees Celcius. Time to cool down to body temperature was 37 s. Infrared lasers simultaneously sealed and bisected blood vessels, with treatment times comparable to, and temperatures and cooling times lower than conventional devices.

Over time the facilities management industry has experienced growth, which has led to a notable surge in the number of professionals ready to undertake the available entry level and senior level positions. These professionals have different personality characteristics and behavioral tendencies. Determining their personalities offers a comprehensive picture of their social interactions, interests, feelings, and workplace behaviors. Therefore, the aim of this research is to determine if there are differences in personality profiles of entry level professionals and senior level managers within the facility management industry. To accomplish these goals, the study will employ various methods including the HEXACO personality inventory, assessments of Emotional Intelligence and Q-DiSC behavioral diagnostics that offer insights into an individual's priorities and preferences in the workplace. These assessments have the potential to identify and explain the personality traits and unique characteristics of people. The survey will be disseminated to facility management professionals throughout the United States. Previous studies in the construction industry have shown that senior leaders tend to show strong personality skills in leadership, fearfulness, and higher emotional intelligence as compared to entry level professionals. It is anticipated that this research study will provide similar results due to the similarities between the construction and facility management industries. The findings will not only contribute to a deeper understanding of the diverse traits exhibited by facility managers but also offer practical insight for enhancing team dynamics and optimizing managerial effectiveness in the field. This research holds significance for both academia and industry, providing valuable implications for human resources and professional development within the facility management domain. 

Domain adaptation is a significant obstacle in developing robust, generalizable AI models for medical imaging applications. The challenge is more pronounced in retinal imaging due to the limited training dataset and diverse range of diseases. This paper presents a novel FL framework that leverages recent advances in a self-supervised large language modeling called FedMIM. This work used this framework for classifying age-related macular degeneration (AMD) using optical coherence tomography (OCT) images. We utilized retinal OCT images from three online datasets: Kermany et al., Srinivasan et al., and Li et al., containing normal, AMD, and other diseases (DS1, DS2, and DS3, respectively). We created a simulation scenario using data from multiple clinics (FL nodes). Each node initially trained its model on its local dataset and then shared the model weights with a central server. The server aggregates the model parameters and redistributes updated global weights to each node, completing one round of training. We repeated this process for 10 rounds. We used a masked image modeling (MIM) encoder with 30% masking to extract meaningful feature representations of the input. We pre-trained the encoder (on all OCT data regardless of its label). Then, each node used the binary classification head to utilize FL-aggregated MIM (FedMIM) as a feature extractor and fine-tuned the model on local data. Both pre-training of the MiM encoder and fine-tuning took one training epoch. Finally, we evaluated the fine-tuned model on the corresponding test set. We compared our results with a baseline model without MIM pre-training. The FedMIM model outperformed the baseline approach with an increase of at least 5% when evaluated using the area under the curve in DS1 and DS3 and performed fairly close on DS2, suggesting that MIM models have better generalization capabilities as they can extract strong representations from the input data. 

Organic-inorganic hybrid materials provide unique properties and have several advantages that are not available in either organic or inorganic materials, such as a wide range of tunable optical and electronic properties, lattice-matching flexibility, improved processability, and low-cost fabrication. However, they often suffer from poor long-term stability and crystal structure disorder. β-ZnTe(en)0.5 is a member of II-VI-based organic-inorganic hybrid nanostructures, exhibiting a uniform and fully ordered short-period superlattice structure without physical and chemical fluctuations [1]. The thickness of the inorganic sheets is comparable to 2D materials, and the structure can be viewed as a periodically stacked 2D material. The exceptional long-term stability of β-ZnTe(en)0.5 make it promising for (opto)electronic applications [1, 2]. Space-Charge-Limited Current (SCLC) measurement is a convenient and simple yet powerful experimental two-probe electrical measurement method to investigate charge carrier transport properties (e.g., mobility), electronic trap concentrations, and energy distribution of carrier trapping states of materials (i.e., insulators, organic semiconductors, and hybrid materials). We apply a Space-Charge-Limited Current method to determine the carrier mobility of β-ZnTe(en)0.5 along different symmetry axes. Mott-Gurney (MG) model [1] is often used for interpreting SCLC behaviors. Along the organic-inorganic stacking direction, the mobility is in the order of 10-3 cm2/(Vs) from both pristine and 15 years old samples, and in the plane parallel to the inorganic sheets, the mobility is anisotropic, in the order of 10 - 100 cm2/(Vs). β-ZnTe(en)0.5 showed long shelf life (~15 years), which is a unique property in hybrid materials anisotropic mobility, and higher mobility values than conventional organic semiconductors, but slightly smaller than inorganic semiconductors attract optoelectronic applications.

This paper delves into a comprehensive exploration of the efficacy of a 275 nm UV-LED system in deactivating two pivotal bacteriophages, MS2 and Phi6, serving as surrogates for non-enveloped and enveloped viruses. The study involves the development and implementation of a precise 275 nm UV-disinfection system, emphasizing meticulous control over UV exposure parameters, including dose, irradiance, and exposure time. Notably, the Phi6 bacteriophage exhibited a remarkable ~1.26 log reduction factor, translating to an impressive ~94.50% reduction in viability, achieved at a UV dose of 32.81 mJ/cm². Similarly, the MS2 bacteriophage demonstrated a log reduction factor of 1.16, resulting in a 93% reduction at a UV dose of 26.82 mJ/cm². These findings underscore the substantial potential of the designed 275 nm UV-LED system as a targeted and versatile disinfection technology, with significant implications for practical applications. Additionally, the exploration of integrating solar cell technology in future research endeavors holds promise for the development of self-sustained and renewable disinfection systems.