Project Summary

The PSI Beta project is fully aligned with the SIF Beta Challenge for Data and Digitalisation, and successful project completion will deliver the next generation of user driven digital products, services, and processes. The project will create a predictive safety model in the gas sector and ready to take to the wider energy sector and utility sector globally, aligned with the Beta challenge phase.
According to HSE annually released statistics, at least 10,000 working days were lost to injury in the wider utility sector in the 21/22 financial year, with the estimated cost of fatal and non-fatal injuries more than £160m. The PSI has a clear and direct target to prevent the occurrences of fatal and non-fatal injuries, which will reduce the cost of operating energy networks, a direct objective and aim of the SIF challenge for Data and Digitalisation.
FYLD has become the tool of choice to manage safety and productivity in the workforce at SGN and has delivered a 20% reduction in incidents and injuries alongside annual financial savings of c.£4.5m. We forecast the safety improvement opportunity from successful PSI completion to be well beyond these outcomes, using predictive analytics to identify which workers and activities will have a safety incident and push teams to intervene and respond before they occur. FYLD's vision is to assist every fieldworker to take corrective actions and put unsafe conditions right in real time.

n the Alpha phase, FYLD and SGN proved it was possible to accurately quantify risk scores in real time and prompt a preventative or mitigating action, deploying the machine-learning model in a proof of concept. The model drew on 3 different live-data inputs, delivering an accuracy of 57%. We demonstrated that the model accuracy was improved through an increased number of data sources, noting that applying the live weather to the model increased accuracy alone by 4%.
Our problem understanding also grew in the Alpha phase with respect to the method of surfacing interventions. In our Alpha submission we targeted building and deploying a control suggestion, however during our governance sessions, we remained agile and built the capability to push high risk notifications - enabling an AI powered human intervention. The outcome was positive - we saw a 35%increase in the response rate from field teams in high-risk vs non-high-risk jobs. We hypothesise that we can increase this improved response through iterations, human validation of recommendations, and improved AI powered interventions, which will be delivered in the Beta phase.
We can say with confidence that we are beginning to accurately predict the presence of a safety risk on site and intervene in real time.
We will take this further in the Beta project, iterating the model through greater data sources. We will build the ability to capture and integrate live situational data from local traffic and roadworks, alongside human related factors such as fatigue, voice tone or behaviour changes. We will develop the object recognition to go beyond detecting objects on controls, and research the ability to detect where site set ups are non-compliant and contribute to safety risk.

User Needs & Personas

• Fieldworkers face a reduced capacity to perceive risk on site due to overexposure, and differing capabilities mean individual risk tolerance varies by individual. The project will seek to address the lack of access to data of historic incidents or safety events, or the inability to draw a link between those safety events and risk factors that may be present
• Field team managers - expectations exist to interact in many risk assessments, but time demands mean that this cannot always be immediate. By enabling managers to focus their attention and prioritise sites identified as high risk from live data inputs, we can target a second set of eyes where it is needed most and shift away from ineffective sampling techniques
• Senior and safety managers - further removed from site activities, senior and safety managers need to have confidence in, and the ability to visualise, risk management. Creating risk quantification and visibility via interactive dashboards, and the ability to performance manage the associated mitigation, or be alerted when risk hits unacceptable levels, are key drivers for this persona group

FYLD are best placed to assist SGN and bring this solution to market through:

1. A high-performing team with experience launching and maintaining AI/ML products in the remote field service industry
2. A demonstrable history of realising significant cost savings for utilities companies by deploying innovative solutions
3. Existing technology and datasets that can be built upon
4. In-house experience and expertise in change-management required for digital transformation, specifically within safety and productivity of utility companies, at scale

Innovation Justification

HSE data shows reduction in safety incidents in the Utilities Sector has plateaued for nearly 10 years, and in fact rose by approximately 6% in 2021-22. The counterfactual solution of 'Do Nothing' has ceased to deliver further reduction in fatal and non-fatal injuries.
Our Alpha project saw the delivery of the prediction model. We hypothesised that by accurately predicting high risk sites before an incident occurred, and prompting an intervention ahead of time, we would build a product with the capability to significantly reduce fatalities and non-fatalities across the industry. We successfully demonstrated we can accurately predict safety events to an estimated accuracy of 57%; and we successfully evidenced we can improve the response rate to high-risk jobs by increasing the intervention rate by 35%.
We will take this further in the Beta project phase by increasing the sources of data to further develop the accuracy of the predictive model and iterate our intervention to deliver tailored and specific learning from previous safety events. This will push a tailored intervention directly into the hands of somebody working in similar conditions that has the potential to result in an injury.
The PSI model is a giant step forward in innovation in the sector. Existing products capture learnings from safety indicator events through basic form data capture but are unable to share the learnings to where they are needed most, by the fieldworkers most vulnerable to incidents and injuries. By automating learning from indicator incidents and applying them across industry when reviewed in conjunction with real-time data inputs, we are taking a leap forward from industry standard of manual reports, manual incident reviews with learnings shared via safety stand-downs and written communication.
Our Beta project will take large steps forward in developing the AI predictive model. We will incorporate human factors such as fatigue levels into the model and undertake industry leading research on the validity of including voice tone and pitch changes as a demonstrator of human behaviours. We will also further the object recognition training, and research the ability to detect non-compliant sites in line with safety standards as an input into the model - a first for the industry to have computer vision analysing sites for non-compliances.
Rather than relying on fieldworkers to manually report safety events in the hope of preventing them next time, we believe technology is better placed to:

• Capture high-fidelity data
• Recognise patterns in the data
• Analyse growing risk factors across several safety related factors
• Make Interventions in real time automatically triggered when thresholds are met

Commercial readiness levels (CRL) of the PSI model are currently at level 6, in Product/ Solution Optimisation phase. The MVP is built with good early usage data. Our forecast CRL post Beta phase is an 8, with customer validation achieved and the PSI model being introduced to the market.
Integration readiness levels (IRL) is currently at level 6. Integration is achieved with data sources into the model, and FYLD has a bi-directional open API. Our forecast IRL post Beta phase is 7. An additional API type will be developed to integrate asset information directly into the PSI model from SGNs asset management system.
The scale of the Beta project is fully aligned with the SIF Beta Challenge and the target objectives. The Beta project will take the PSI from a proof-of-concept phase with strong early outcomes, to an optimised model deployed in the gas sector and ready to take to the wider energy sector and utility sector globally, aligned with the Beta challenge phase. The clear and direct benefit of reducing fatal and non-fatal injuries will reduce the costs of operating networks, in line with the SIF target benefit.
This project is not suitable to be funded or managed as BAU from either SGN or FYLD. FYLD has spent 18 months in the market growing our customer base across the gas and utilities sectors, and globally across safety critical industries. The product has demonstrated success, and therefore the business focus of FYLD is to further our client base rapidly. Successful Beta funding approval would be used to grow out a capability within the product which can deliver outcomes for the sector but does not have proven marketability to fit in-line with FYLD commercial ambitions. Therefore, the commitment of resources for FYLD is too high risk to undertake as BAU.
For SGN, the PSI project aims to deliver a safety innovation outside of our BAU capabilities. SGN is currently investing under BAU to improve safety culture and practices is seeking incremental improvements to bring SGN in line with best practices. This project is seeking to significantly advance the best practices available in industry. SIF also provides an open platform for sharing knowledge and project outputs, allowing for benefits to be shared to the wider industry.

Benefits

Successful completion of the Beta phase will deliver financial benefits through a cost reduction in operating energy networks and wider energy systems. This will be delivered by:

1. Reduction of injuries and incidents
2. Lower cost to capture data and share learnings about indicator events

In addition, successful Beta completion will deliver a new to market product and process. Use of predictive safety models to manage and mitigate risk is new to the Utility sector in the UK, with no established predictive models present in the market. This will deliver a core benefit in line with SIF target benefits of:
3. Improved risk visibility across network operations
Within our CBA options in the attached project management tracker, we have run 2 CBA options. Option 1 is full-scale deployment at only SGN within the first 2 years from project commencement, option 2 is deployment to wider industry assuming a 30% uptake from the gas sector only. We have not forecast benefits to the wider Utilities network, including the water industry and other similar industries, as this is not under OFGEMs remit, however it is reasonable to assume that similar benefits would be applied in other industries, for example the water industry.
1. Reduction of lost-time injuries
Based on HSE data, the cost of lost-time injuries in the wider utility sector in 2021-22 amounted to more than £160m (HSE data). The CBA baseline scenario shows the estimated cost to the gas sector over £48m. We hypothesise that through accurately predicting safety events and intervening before a fatality or non-fatal injury occurs, we can substantially reduce the occurrences of fatalities and non-fatal injuries and reduce this cost.
In the Alpha phase, we demonstrated 57% accuracy of the PSI model, and we demonstrated a 35% improvement in response rate from workers to improve safety on site or review the site conditions. In Beta, we are targeting achieving at least a 65% accuracy, which is delivering similar accuracy to other market leading prediction models, for example in the medical sector. If the PSI model could predict 65% of incidents before they occur, with a 75% success rate, the potential annual saving could be in excess of £60m per annum when widely adopted across the industry.
2. Lower cost to capture data about indicator events

In the CBA baseline scenario, we estimate the cost to gas industry, if the delta was addressed between actual and actual near miss reports, to be approximately
£7.8m. This doesn't include the cost of reviewing and analysing safety incident events, data and trends, and documenting the learning to work forces via safety communication.
Based on the number of lost-time injuries, in the 21-22 financial year, SGN should have reported over 66,950 indicator events including near-misses, based on the Heinrich and Bird safety triangle, widely accepted in the industry. In the Beta phase, we will develop the PSI solution to near-automate the capturing of indicator events as the accuracy of the AI model develops. PSI will deliver substantial time and resource saving from replacing manual processes with an automated model with human validated data collection i.e., the PSI model will be able to detect when a likely safety incident has or will occur and request the worker to validate this. The assumptions in this saving are modest and do not include the extensive time require to analyse data and share learnings.
3. Improved risk visibility across network operations
In line with the target SIF benefit for new to market processes, PSI Beta will deliver the qualitative benefit of improved risk visualisation through performance reporting against predictive safety analytics. Operational and safety leaders across the sector will have the newly developed ability to report on safety trends with live quantification of risk enabling a snapshot view into an AI quantified risk level across real-time sites and workstreams. The reporting and visibility of response to high-risk safety events has the potential to become the first of its kind in enabling cross-company benchmarking of risk through enhanced prediction-based risk visibility.
We have also identified the following key benefits which align to SIF target benefits which have not been quantified for this project:

• Service strike & incident reduction leading to an increased supply security for vulnerable and non-vulnerable customers due to fewer supply interruptions
• Gas leak and major event reduction leading to a proportional reduction in emissions from major gas leaks, hydrogen leaks, explosions, and other major incidents

Cumulative Discounted Net Benefits CBA Option 1 - SGN 100 % Adoption 1 Year - £0.07m
3 Years - £5.16m
5 Years - £9.91m

10 Years £20.46m
CBA Option 2 - SGN 100% adoption + 30% Gas Industry Adoption
1 Year - £0.41m
3 Years - £7.12m
5 Years - £14.59m
10 Years - £35.80m