Artigo Revisado por pares

Richard Parry-Jones 1951-2021

2021; Royal Statistical Society; Volume: 185; Issue: 1 Linguagem: Inglês

10.1111/rssa.12751

ISSN

1467-985X

Autores

Tim Davis,

Tópico(s)

Advanced Statistical Process Monitoring

Resumo

Richard Parry-Jones (or RPJ as he was referred to by his colleagues) was born in Bangor, Wales and graduated with a first-class honours degree in mechanical engineering from the University of Salford, with a specialism in control theory. He joined the Ford Motor Company, originally as a graduate trainee in 1969, and was hired permanently after graduating in 1973. He enjoyed a 38-year career with the automaker, rising to be the Group Vice President for Product Development, and eventually Chief Technical Officer. In this role, he was responsible for 30,000 engineers worldwide, across all the brands acquired by Ford (including Jaguar, Land Rover, Volvo, Mazda and Aston Martin). Along the way to achieving the most senior position in Product Development within Ford, Richard ran the Body and Assembly plant in Cologne, Germany, which produced the Scorpio and Fiesta models. In this phase of his career, he gained an appreciation of the role of statistics in industry and implemented statistical process control within the factory. Ford’s initiative in this area was influenced by discussions between its chairman Donald Petersen and WE Deming. On Richard’s return to engineering and product development in around 1990, a natural question to ask was ‘if statistical methods can help with manufacturing, could they also be of benefit in engineering?’. At around this time, Ford had developed an extensive training programme called EQUIP (the engineering quality improvement programme), which entailed applying statistical methods to the quality of the engineering activity with a heavy emphasis on experimental design, robustness and response function estimation. The burden of quality control was consequently lifted from being the exclusive responsibility of manufacturing operations. To demonstrate their commitment to the EQUIP programme, senior managers attended the first training sessions so that they were ‘equipped’ to deploy the teaching to their respective teams. Richard was a member of this first training cohort. The first product programme where Richard was able to use his statistical engineering influence from the beginning of development through the production was the original Ford Focus launched in 1999. He was particularly struck by the concept of engineering robustness that he had learned on the EQUIP programme. This included formulating designs that achieved their functional targets despite the effects of sources of variability (called collectively noise factors). He was determined to integrate this thinking into the quality planning of the Focus. In design reviews, he enjoyed drawing a non-linear response function to demonstrate to the engineers that the amount of subsequent transmitted variation to the total design depended on the gradient of the response function at the chosen design nominals. The motivation was to get the engineers to realize for themselves that they were, perhaps, not quite as finished as they had first thought. To help engineers explicitly identify the relevant noises for the component that they were responsible for, the noises were categorized into five types, a classification system that is now widely adopted in the automotive industry and beyond. The Mk 1 Focus was voted Car of the Year in 1999. Above and beyond the accolades the car received for its handling and driving dynamics (both an RPJ speciality), it gave Richard great satisfaction that the car was recognized as being very reliable. This reflected the rigour that went into ‘signing off’ each component against the appropriate set of noises. In 2003, a few years after it was launched, the Mk 1 Focus was the first non-Japanese car for many years to win the prestigious TÜV reliability award in Germany. Over 20 years after it was launched, the Mk 1 Focus is not an uncommon sight on the roads today. The next programme to receive Richard’s full statistical engineering treatment was the Mk 2 Mondeo (the Mk1 having begun development during his stint in manufacturing). The approach to robust engineering deployed on the Focus was further refined by integrating robustness against the five noises into the FMEA. A robustness assessment was formally embedded into all design reviews, an approach to reliability that Ford called ‘Failure Mode Avoidance’. The Mk2 Mondeo just missed out on being awarded 2001 Car of the Year due to a technicality. In 1999, Richard gave the annual engineering and manufacturing lecture at the Royal Academy of Engineering (‘Engineering for corporate success in the new millennium’). This lecture laid out in a superb narrative style with real examples, how Ford was integrating statistical science into mainstream engineering practice (as evidenced by the Focus and Mondeo programmes). In his lecture, due reference was given to the idea of transmitted variation which was credited to SJ Morrison. No lesser statistical luminaries than George Box and Stu Hunter subsequently wrote to Richard saying ‘we were simply bowled over by your lecture; the value of the synergy between engineering and statistics could not have been better expressed’. Reflecting his success at integrating quality using statistical methods into mainstream engineering practice, Richard was appointed by the Ford Motor Company as Group Vice President of Product Development and Quality. This necessitated a move from Europe to the United States. In 2000, the Firestone tyre crisis erupted in the United States, when certain Firestone tyres suffered from tread separation causing vehicles (mostly the Ford Explorer, an SUV) to roll over, resulting in around 300 deaths. The approach of the investigation into the cause carried out under Richard’s leadership was ‘to find the truth, even if it means that the Ford Explorer was at fault’. He was by now well equipped to oversee the statistical effort required to get to the truth. Statistical applications such as response surface experiments and proportional hazard regression analysis were deployed to confirm the root cause of the problem (which was in fact due to problems with the tyre, not the vehicle). When tyres fail catastrophically, they destroy any internal evidence of what might have been wrong with the tyre. But in a perfect example of working at the interface between statistical science and engineering, it was determined that immediately prior to tread separation failure the tyre exhibited higher levels of vibration. With suitably engineered instrumentation, the tyre could be stopped on the test rig immediately prior to its impending failure and its internal integrity could then be established. Without this synergistic approach, the hazard regression analysis, which was the last statistical step in establishing the root cause, could not have been performed. Richard was on record as stating that resolving the Firestone tyre crisis was the toughest engineering challenge of his career. After he retired from Ford in 2007, Richard took his enthusiasm for statistical engineering to Network Rail where he was non-executive chairman from 2012 to 2015. He commissioned and oversaw experiments aimed at optimizing design robustness of the electrification of the Great Western Mainline (GWML) from London Paddington to Cardiff. The engineers had thought a priori that their design was finished, only to find after conducting the experiments, the tension in the contact wire had to be increased from the original specification to achieve acceptable robustness of the overhead line equipment. This in turn required deeper foundations for the gantries than had been originally specified. Echoes from the Ford Focus programme from 20 years previously did not go unremarked! Richard was honoured by several universities, including Loughborough (where he was a visiting Professor and former Pro-Chancellor), Warwick and his hometown university, Bangor. He held fellowships with the Royal Academy of Engineering and the Institution of Mechanical Engineers. He was made an honorary fellow of the Royal Statistical Society in 2004, reflecting his influence on statistical engineering. He was also awarded a CBE in 2005 for services to the automobile industry. Despite being ‘retired’, he ran his own engineering consulting company and took on other appointments, such as chairman of Yorkshire Water and non-executive director of GKN. He also chaired the Automotive Council UK, under the auspices of the Department for Business Innovation and Skills. At the time of his death, he was chairman of Marshall Motor Holdings plc, and chairman of the Faraday battery challenge, aimed at developing battery technology for electric vehicles. In 2015, Richard suffered a stroke and a heart attack, and allegedly slowed down a little. He devoted more time (together with his wife Sara) to running his award-winning Spa and Lodge, situated opposite his magnificent home on the Barmouth estuary. He designed much of the interior and infrastructure of the lodge himself. Reflecting Richard’s commitment to the environment, the lodge was not just Carbon neutral, but Carbon negative, as it consumed more Carbon than it emitted. It was while working in the Lodge grounds on his tractor that he suffered another heart attack, which resulted in his death. As the news spread, his many colleagues and friends around the world communicated via email: the subject line just said ‘Richard’. That was all that was required. He is deeply mourned by all who knew him and worked with him.

Referência(s)