TECHNGI-CDT, Centre of Doctoral Training
This newly established centre – jointly funded by WTW Research Network and Loughborough University – is offering full fee PhD scholarships for research on the adoption of digital technologies in Industrial and Commercial Property Insurance. It builds on the earlier co-operation between WTW Research Network and Loughborough University in the now completed TECHNGI research project on technologies and next generation insurance.
The Centre will run from July 2022 until Sept 2026, with a total of six students joining in two cohorts. The research will be conducted in close co-operation with the WTW Research Network and their parent, the global insurance broker WTW, with each student having an industrial supervisor or advisor as well as academic supervisors, access to relevant data through WTW and engagement with practitioners throughout their studies.
The research topics together form a cross-disciplinary program of research examining foundational questions about the application of digital technologies in industrial and commercial property insurance. We envisage this initial research as the incubation of a broader industry effort to adopt digital tools in insurance risk assessment and operations (e.g. on agreed benchmarks for consistency, independently contributed open-source code, the development of software tools). The young researchers who graduate from this program will be ideally placed for playing leading roles in the digital transformation of industrial and commercial property over the years ahead.
The Centre itself will allow students to pursue their individual research topics under the umbrella of a broader academic- industry student community, all addressing the challenges of digital transformation in property insurance. This community will bring the research students and their supervisors together, typically virtually but also in face-to-face meetings, with a range of other researchers and practitioners engaged in related work. The PhD scholarships also include funding for relevant field work costs, scholarly engagement through conferences and external training. The three students in the first cohort will be registered in Civil Engineering, Computer Science and in Business and Economics and offered cross school supervision so they are exposed to different disciplinary perspectives.
PhD descriptions for first cohort in 2022
Ontology Digital Twinning for Knowledge Modelling and Fusion in Building Risk Assessment and Insurance
Built on the initial work of ‘Applying AI to building blueprints for insurance risk assessment (AURIE)’, this project aims to develop an ontology digital twinning approach to construct the domain knowledge graphs for improved risk assessment and underwriting capability. It will:
- develop an ontological knowledge framework from CAD drawings/BIM models (building ontology), safety codes (safety/risk ontology), and sensor data (occupant behaviour ontology) for risk knowledge modelling of buildings.
- develop a multi-source knowledge fusion method based on the ontology frameworks for a uniform domain knowledge model.
- develop arrangements for automatically querying the uniform knowledge model for risk assessment of buildings.
- explore the practicalities, including software or tool development, for implementing these techniques using the blueprint and other information for commercial and industrial properties available to WTW and WTW clients.
Its ultimate ambition is to transform property insurance risk analysis and underwriting from present arrangements employing extensive manual processing into an automatic and reliable process using ontology digital twins.
AI-based Autonomous Building Environment Mapping and Fire Asset Detection for Resilient Risk Management and Insurance
The aim of this PhD project is to develop AI and computer vision technology to automatically detect fire infrastructures and assets (e.g., fire extinguishers and emergency signs) and other important objects related to fire risks, emergency evacuation and rescuing in buildings. Novel AI technology will be developed to automatically build layout maps of buildings (e.g., rooms, doors, corridors), presenting semantic data of above extracted fire-risk object information. The developed system can also have the capability to update the mapping and identify any changes since the last survey. The object detection and building mapping will be based on the information in video images captured by cameras.
In particular, the state of art deep learning and deep neural networks will be investigated for visual detection and pattern recognition of objects and layout. Real-time location and mapping using visual landmarks and features will be developed. Models will be compared and validated using real-world building data. Transfer learning techniques will be deployed to deal with limited image samples during the AI model training process. Initial data collection and experiments will be carried out at building at Loughborough University campus. Further data collection, experiment and tests will be carried out at different building types in collaboration with the WTW Research Network.
Technology, competition and risk transfer in commercial and industrial and property insurance
Industrial and commercial property insurance, like other forms of insurance, is a market for risk transfer. This project will use market data provided by WTW and other data sources, including insights obtained from research conducted by other students in the mini-CDT, to address central questions on the efficiency and competition in this market and the potential impact of digital technologies on risk analysis and risk transfer. The research will initially examine the magnitude and determination of the risk premium, the excess of insurance premia over actuarially fair expected pay outs. This work will be pursued both from a theoretical perspective, capturing the role of information asymmetries, and empirically using WTW and other data. The research can also investigate the dependency of the risk premium on property specific risk information and market factors and the impact of technological innovation, including automation of risk data, on the setting of risk premia and on market competition. The development and execution of this program of work will be conducted in close co-operation with WTW staff and WTW clients.
PhD topics for the remaining three students in the 2023 cohort.
AI-based 2D-3D joint modelling and enrichment of as-built BIM for property insurance
Building Information Modelling (BIM) has been increasingly adopted as a shared information resource and knowledge base for building and facility management. Property insurance companies can also leverage such information for pricing and underwriting. However, there are two major challenges for accessing as-built BIM data, including (1) the lack of as-built BIM models for many existing structures (e.g., old properties), and (2) deviations between BIM models and actual builds. Laser-scanning and photogrammetry methods have both been developed for as-built BIM reconstruction and enrichment, but the performance of such methods have been degraded due to the increasing complexity of buildings, such as occlusions and specular surfaces.
This research aims to address these challenges through developing a 2D-3D joint modelling and enrichment approach for as-built BIM based on the integration of laser-scanning and photogrammetric point cloud data. It aims to combine the advantages of both methods to generate a complete and accurate BIM model, and finally update this model according to the actual property states. This research mainly consists of the following four interconnected phases:
- Laser-scanning point cloud modelling.
- Photogrammetry-based modelling.
- Integration and registration for complete BIM model.
- BIM model updating and enrichment.
Seeing extreme winds: Using video clips to improve wind hazard estimates for risk mitigation
This PhD project will contribute to solving the real-world challenge of estimating wind hazard for industrial/commercial property by applying image analysis techniques to short video clips of trees. You will have the opportunity to conduct an exciting blend of observational work (field & laboratory), physical modelling and image analysis. We envisage a campus-scale study linked to the university’s weather station, a variety of image collection (initially with a mobile phone, then exploring the intriguing possibility of quasi-continuous CCTV camera data), and possibly wind tunnel experiments. Initially, work will build on data from a pilot study and existing particle-tracking code to create a local (micro-scale) map of wind hazard. The final aim is to create a proof-of-concept tool for high-resolution, low cost mapping (i.e. high/low hazard areas) without installing specific technology (wind meters) enabling site-specific recommendations on mitigation (e.g. tree planting as a shield).
Knowledge management in industrial and commercial property insurance
Agreement on the terms for insurance of industrial and commercial properties is based on assessments of potential loss by risk engineers arising from a variety of hazards together with the exposure and vulnerability of the property to the materialisation of each hazard. While standardised quantitative methods are available for modelling natural hazards such as flooding, exposure, vulnerability and the risk analysis of other hazards such as fire, explosion, crime and are based on the judgement based assessments of experienced professionals using a mix of qualitative and quantitative methods. A central challenge in applying technology to property risk assessment and insurance is developing a digitally supported management of all the supporting professional knowledge and practice that underpins these risk assessments.
The aim of this project is to understand the complexity of the knowledge management problem in the risk assessment of commercial and industrial property. It will review the various knowledge management practices at the different countries, and design knowledge management and digital transformation strategies to help WTW to improve the dispersed knowledge management processes used in these risk assessments. It can also resolve problems and inconsistencies in the management of knowledge in property risk assessment.