Impact case study database

Background to RUMM Ltd and collaboration with USW

RUMM Ltd [5.1] was a new company launched in South Wales by engineering lecturers, Dr Steve Lloyd and Mr Steve Thomas, as a spin-out company from the University in 2005 following a collaborative project with local companies to remotely monitor their energy consumption. The unique service offered by RUMM was aimed at medium-sized businesses, benchmarking their energy usage against critical parameters and the subsequent application of statistical process control techniques to reduce their energy bills and their carbon footprint. These savings aided the businesses in achieving targets in the wake of deregulation of the energy supply infrastructure in the UK and the increased political drivers for carbon reduction.

RUMM’s technology provided a remote assessment of each client’s energy use, typically using between 20-60 meters per site that measured energy consumption at key locations. Time series data from each meter was transmitted to RUMM’s server in 30-minute intervals where it was automatically analysed using RUMM’s proprietary software package, Internet Based Analytical Software Suite (IBASS). IBASS provided the client with a visualisation of energy consumption and targets of energy consumption using data analytics. Automatic alarms were sent to the users whenever a meter identified an abnormally high energy use. For example, an alarm would be triggered by a machine being left on outside of normal operating hours. However, the analysis of the time series data in IBASS, which was central to the RUMM model’s ability to detect inconsistent behaviour was identified by the company as basic; it simply compared energy usage in successive half-hourly intervals in a rudimentary manner and lacked any sophistication to deal with changes in demand due, for example, to seasonal differences. To generate the greatest possible savings in energy consumption and carbon emissions, the company sought a collaboration with the University to improve its energy management methodology.

Pathway to impact

In 2006 RUMM approached Al-Madfai to develop the advanced statistical tools they needed to improve the analysis of their clients’ energy and utility data, and management of usage. Following the identification of the presence of multiple seasonalities and heteroskedasticity in RUMM’s clients’ data, Al-Madfai and the company submitted a collaborative bid for funding from the EPSRC for an Industrial CASE PhD Studentship [3a] (titled Detecting Irregular Energy Consumption Through Analytical Techniques (DICTAT)), which was awarded in 2008. Akinbami Akinwale was recruited to the project as the research student, to build on Al-Madfai’s own PhD work [3.1] to develop an analytical technique that could be used as an energy management tool in detecting irregular behaviour of a time series [3.5]. To achieve this, the Hierarchal Profiling Approach (HPA) originally developed by Al-Madfai, Ryley and Ameen and extended from 2002 by Al-Madfai and co-workers, was extended further. Process control techniques were developed to detect abnormalities extracted from RUMM’s time series data that identified “non-standard” consumption profiles, while accounting for the seasonalities and variabilities embedded within the data ( https://youtu.be/xs1YboaQhT8). This enabled development of a remote monitoring and alarming system for the detection of irregular consumption patterns. With research outcomes having contributed to continuous upskilling of RUMM staff and to upgrades to IBASS, the statistical tools utilising HPA completed by Al-Madfai and co-workers formalised the toolkit as a key component of RUMM Ltd’s intellectual property and its competitive advantage from 2013 onwards. The intellectual property in the results of this research project were assigned to the company on the 31st March 2015 in a legal agreement [5.2a] to allow the company to incorporate them into IBASS Version 6 [5.2b] as a key part of a commercial deal with a major energy provider RWE npower. RUMM continued to deliver services supported by IBASS until 2019 when the company was fully subsumed into the new parent company.

Impact: Acquisition of RUMM by RWE npower

By the end of year trading on 31^st August 2014, RUMM Ltd had an annual turnover of approximately £1.3 million, generating £200 000 pre-tax profit [5.3]. On 31^st March 2015, RUMM was purchased by RWE npower Ltd [5.4], known as one of the “Big Six UK Energy Suppliers” and a subsidiary of E.ON UK from January 2019. The statistical analysis and control tools developed by Al-Madfai and co-workers formed a central component of the company’s intellectual property. To quote Mr. Steve Thomas (the commercial director of RUMM Ltd) on its sale to npower, “If Hasan [ Al-Madfai] wasn’t there, it couldn’t have been done” [5.5].

At the point of acquisition, RUMM was reported to have saved its customers £43 million in energy costs, reduced energy consumption by 600 million kWh with a concomitant reduction of carbon emissions by almost 300 000 tonnes [5,1, 5.6]. RWE npower paid [text removed for publication] for RUMM [5.7], and immediately began rolling out RUMM’s energy-saving approaches to its customers in the second quarter of 2015.

Impact: ongoing benefits of RUMM to RWE npower and its UK clients

Following the acquisition of RUMM by RWE npower, RUMM became part of npower’s Business Solutions suite of packages, under its “Energy HQ” umbrella [5.8], which continues to be offered to its customers. Later in 2015, RWE npower became the first power utility to be awarded the Carbon Trust triple standard for significant reduction measures across energy, waste and water, with the acquisition of RUMM cited as one of “[two] key developments that led to certification” [5.9]. Npower’s Low Carbon & Behaviour Change EMS Manager reported of RUMM a year after acquisition: “It’s already produced clear and proven benefits for firms, big and small, saving businesses we have worked with over £26m” [5.10], demonstrating the value to npower of the IP purchased. It was anticipated by npower’s Chief Executive that by rolling out RUMM’s technology across its customers, £4 billion could be saved from the UK’s annual fuel bill for businesses across its 23 000 commercial customers [5.6]. Npower targeted savings of between 3-15% for each of its customers through its focus on “RUMM’s behavioural approach to energy” [5.10].

Impact: upskilling of employees

A key philosophy of RUMM Ltd was the upskilling of its employees. Based in Ystrad Mynach, Caerphilly County Borough, RUMM’s 30 employees [5.6] were predominantly drawn from local communities that display poor educational achievements and high overall deprivation, and the majority of staff were appointed to positions requiring high technical skill sets. The 2019 Welsh Index of Multiple Deprivation ( Welsh Index of Multiple Deprivation (WIMD) 2019: results report (gov.wales)) identified Caerphilly County as the 4th most education deprived and 5^th most employment deprived local authority of 22 in Wales (by percentage of Lower Layer Super Output Areas in the 50% most deprived such areas). As a result of the development of the HPA method and its incorporation into IBASS, a number of RUMM’s staff were upskilled in their statistical knowledge [5.5], adding further commercial value to the company (as evidenced by RUMM Ltd’s sale price [5.7]) and reducing both employment and educational deprivation in the region. Further, a former research student working within RUMM subsequently became a director, and post-sale developed Blue Sky Equity Ltd (see below), exemplifying, through the collaboration, the research-driven development at USW of entrepreneurship skills of students and staff.

Impact: investments from profit generation

Shareholders’ income from the sale of RUMM Ltd has led to further new company starts and investments in start-up companies from 2015 to 2020. One example is the bio-technology company Blood Line Ltd [5.11], founded in 2017 by two of the ex-RUMM directors to develop technological aids for people with diabetes, making use of statistical techniques to model relationships between food intake, exercise, and key biometric parameters. Further examples include investments in the four companies by another ex-director of RUMM, including Blue Sky Equity Ltd which was established to specifically invest in start-up companies [5.12].

Improved weather forecasts and climate predictions have a significant effect on the economy and on public safety, ranging from better use of energy reserves to advanced warnings of natural disasters [5.1]. Accurate global modelling systems rely on accurate numerical methods to solve the corresponding equations. The development of these methods, along with their careful assessment and evaluation, is essential for improving forecasting capabilities and having confidence in the model results. Kent’s new approach has significantly improved the accuracy, stability and computational efficiency compared to older schemes and enhanced confidence in the predictive capabilities of the models.

Novel modelling system developed by Kent leads to improvements in NASA’s technologies

Kent designed and developed a new tracer transport scheme for NASA’s GEOS-5 (their operational model for weather and climate prediction) linear model. This scheme became the default option in NASA’s GEOS-5 linear model in 2018 [5.2, 5.3] and, as a direct result, showed immediate improved performance over the previous model version as confirmed by the Data Assimilation Lead for the Global Modeling and Assimilation Office at NASA:

“The improvements in the linearized GEOS model as a result of the implementation of the dynamical core with HK [Holdaway and Kent] scheme were very impressive, and significant compared against gains seen for implementation of other components.” [5.2]

[Text removed for publication] the NASA model improves their forecast ability, and thus impacts their model and data users. NASA’s work on global weather and climate predictions is used to inform industry, government and the public in the both the United States and world-wide [5.1,5.5], with vast volumes of data being open-access (e.g. https://earth.nullschool.net/). Within the Global Modeling and Assimilation Office of NASA alone, 20 missions currently exploit aspects of the GEOS system which in turn impact on further campaigns [5.6]. Since NASA share their climate data and models with other centres, such as the National Oceanic & Atmospheric Administration (NOAA) [5.7], improvements to their model mean improvements to the quality of this data world-wide and allow NASA to achieve its key missions. The GEOS linear model is also being used by the Joint Center for Satellite Data Assimilation (JCSDA) whose member organisations include NASA, NOAA, the U.S Navy and the U.S. Air Force. JCSDA is developing JEDI, a data assimilation system, to become a world-leader in environmental analysis and prediction. The use of Kent’s method in GEOS-5 has led to [text removed for publication] the JEDI system as stated by JEDI’s Development Team Research Member [5.3]:

“The HK scheme is by far the most efficient transport scheme available in the GEOS linearized model, making it very attractive for use in JEDI…. [The HK scheme] offers excellent matching between the nonlinear transport and the linearized transport and the best performance compared to other forms of linearized advection tested in GEOS.” [5.3]

Following the success of Kent’s scheme within the GEOS system, NASA is currently implementing another method of Kent for other variables within the linear model of GEOS-5, such as vorticity, temperature and pressure [5.3, 5.4a].

Creating a worldwide standard for test cases in atmospheric models

In 2016, Kent presented his research at a DCMIP workshop, providing at least 9 atmospheric modelling groups around the world with attendees from 5 different continents with a new set of standard idealised test cases that assess the dynamical core’s performance when coupled to simplified physics routines. The groups that took part in the workshop are truly international and drawn from a mixture of higher educational institutions and governmental funded agencies including:

• National Center for Atmospheric Research, Boulder, CO, USA.

• Laboratoire de Météorologie Dynamique, Institut Pierre-Simon Laplace (IPSL), Paris, France.

• Geophysical Fluid Dynamics Laboratory (GFDL), Princeton, NJ, USA (the developers of the dynamical core used in NASA’s GEOS-5 model).

• European Centre for Medium-Range Weather Forecasts (ECMWF), Reading, UK.

• Environment and Climate Change Canada (ECCC), Dorval, Québec, Canada.

• Max Planck Institute for Meteorology, Hamburg, Germany.

• Deutscher Wetterdienst (DWD), Offenbach am Main, Germany.

• Naval Research Laboratory, Monterey, CA, USA.

• Japan Agency for Marine-Earth Science and Technology, Yokohama, Kanagawa, Japan.

The Scientist attending on behalf of the European Centre for Medium-Range Weather Forecasts, an independent intergovernmental organisation supported by 34 member states, noted that:

“The test cases developed by Dr. James Kent of the University of South Wales and used at the DCMIP 2016 workshop are becoming a worldwide standard in evaluating dynamical cores in the atmospheric modelling community.” [5.8]

These groups used the test cases to assess and give confidence in their different models and used the results to guide their model design decisions [5.8, 5.9]:

“… by using these tests, we have great confidence in the performance of IFS-FVM and hence are continuing to develop this technology.” [5.8]

This increased model confidence and improved design decisions impacts both the model data itself and external users. For example, DWD provide services to numerous organisations including the German Government, private weather forecast agencies, TV stations, as well as more than 20 national weather services worldwide [5.10a]. Following DCMIP-2016, DWD used Kent’s test cases to assess and verify their ICON model dynamical core giving them, and hence their stakeholders, confidence that the ICON model is performing well and producing accurate forecasts. Indeed, DWD acknowledged that Kent’s test cases have made valuable contributions to their ICON model [5.10b]:

“… we discovered that the ICON dynamical core erroneously conserves total mass instead of just dry air mass…. we have designed a new approach to remedy the conservation problem….” [5.10b]

In addition, using his knowledge of test cases from DCMIP along with the requirements of the tracer transport linear model from NASA, since 2017 Kent has also created a new set of idealised test cases specifically for linear models. These tests have subsequently been used on NASA’s GEOS-5 linear model, which again highlight the improvements made by Kent [5.4].

Kent’s research has therefore directly impacted multiple national and international weather and climate modelling groups, which make use of his modelling systems. By enhancing their performance and services, his research has also indirectly impacted these groups’ extensive partner organisations and customers worldwide.