Saturday 15 December 2012

Offshore Storms, Statistics and Oil Rigs


(This is my personal blog, views are my own and not those of my present or past employers)

Do you want to know how statistics can help to improve the lives of people working on an offshore platform? If you do, then read the following case study, it is based on a true story. The magnitudes I quote are fact, but for confidentiality reasons the names are fictitious.

It was a cold and wet December day in Aberdeen as Tom and Ray boarded the helicopter. They had completed their journey to the offshore platform numerous times before. The training they had received many years ago at the Robert Gordon Survival Centre had stuck in their mind as they donned their life jackets – what to do in case of abandonment, how to handle a life raft, and fire fighting techniques. On the journey to the platform the noise is as deafening as usual and no-one can or wants to talk as the vehicle makes its journey to the middle of the North Sea. Suddenly, after several hours travel, and through the rain squalls, they could see their destination. They issue a huge sigh of relief. “We’ll be down soon”, says Tom. “Yes” says Ray. A flame pours out of the top of the flare stack and the helideck[1] looms closer as they make their descent.


Oil Platform and Helicopter.
Photo: NASA, via Wikimedia Commons.

After a routine first day, it’s their second night on the rig. The two friends have just finished their evening meal and are looking forward to a film. After a rest they need to be back on shift at seven the next morning so they need some relaxation time. However, it is clear to Tom that a storm is developing because he has some knowledge of meteorology and oceanography. He knew that the swell[2] had disappeared completely, and it had been replaced by wind-sea[3]. Tom looked out and he could see some of the steepest waves he had ever seen. “Ray, it sure is a rough night out there, it looks like a wall of water coming our way. The waves are probably over 20 metres in height”. At that moment one of the operators came in to see the film and announced that the wind[4] was gusting to over 40 metres per second.


North Pacific storm waves as seen from the M/V NOBLE STAR, Winter 1989.
Photo: National Oceanic and Atmospheric Administration (NOAA), via Wikimedia Commons.

Sometime in the night both Tom and Ray thought that they heard a loud bang, but they awoke as usual feeling refreshed and ready for a new day. It was not until they sat down to breakfast that they heard how bad it had been overnight. The weather had by now improved and the wave heights had declined. One of the structural engineers had been to a lower deck, some distance above the sea level, and had seen the damage. Whilst he thought that it didn’t affect a critical part of the structure, he of course recognised the risks and raised the necessary paperwork to alert the offshore company.

After finishing their breakfast the two friends made their way to the spider deck where they were carrying out their work for the day. At this level, near the sea, it can be a dangerous and intimidating place. Tom became reflective. “Ray, I heard that there was a technical guy working with structural engineers at a university. They apparently collect information on the wind and waves from this structure”. “Yeah, I’d heard about that as well. What does he do again?” Ray replied. “I think he’s a statistician“. Said Tom.

Two weeks have elapsed and Tom and Ray have finished their turn on the rig. It’s a beautifully calm but cold day. The sun is shining as the helicopter departs from the helideck for the mainland. “Big game on Saturday, Tom”. “Will you be there ?” “Not sure Ray, I am taking my lad back to university for the new term.” At that moment they stopped talking because of the noise from the helicopter.

So how can statisticians help people, like Tom and Ray, working offshore ?

We statisticians love our data, and it’s the numbers behind our data that make us tick. Some of us might be interested in GDP figures, others drool about environmental statistics, and these days a few of us even get really excited about the difficulties of gathering and crunching terabytes [or should that be yottabytes?] of big data. I suppose you could say that I am in the middle category but I am keeping my eye on the other two.

Offshore companies are interested in fully understanding the environment from a business point of view, and of course from a health and safety perspective. With this in mind, a university helped an offshore company install some of the latest equipment on a platform in the North Sea (including a wave meter, an anemometer and a current meter). The information was sampled by these meters at frequent intervals, so “virtually continuous” data was gathered from these meters. This was then transferred to the university research department for analysis.

"In physical science the first essential step in the direction of learning any subject is to find principles of numerical reckoning and practicable methods for measuring some quality connected with it. I often say that when you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely in your thoughts advanced to the state of Science, whatever the matter may be." Lord Kelvin Lecture on "Electrical Units of Measurement" (3 May 1883)

As a researcher in the department I was responsible for maintaining the offshore equipment on the platform, and for building statistical models that described the environment. So, like Tom and Ray, I embarked on a trip to the platform to calibrate the wind, wave and current meters. I wasn’t there for two weeks on, then two weeks off, and fortunately for me it was very calm at the time. But I did at least get a feeling for the offshore environment and the working conditions.

In a particular storm I had anecdotal evidence that the conditions were of a magnitude that hadn’t previously been experienced and the offshore industry required confirmation of this anecdotal evidence. During the above storm there had been a rapid change in conditions which seemed to have resulted from the passing of an occluded front. In particular, this had the impact of producing high asymmetric waves with steep wave fronts, and these large waves came in groups with few lulls in-between them. The storm peak, due to the wind-sea, was supplemented by coincidence with a high tide, due to the lunar cycle. In the research department I carried out a frequency domain analysis of the time series of wave data, and this informed me that there was significant energy in the waves. This energy increased as the storm intensified.

More generally, the work in the research department was statistical in nature (as opposed to a study of a particular event). For this to be successful a large amount of data, collected at frequent intervals, had to be sampled and analysed. Nowadays, there is so much focus on big data but it’s worth remembering that to make meaningful decisions that help in practice, we statisticians need to perform sampling. In this offshore situation it’s the extreme conditions - the sampled peaks of storms - that make a significant impact to the lives of people like Tom and Ray.

Finally, I am not sure what the two friends are doing now, but from this analysis I do know that the offshore industry has a better understanding of their environment. This research work was funded from grants which were provided by the Engineering and Physical Sciences Research Council (EPSRC). The final reports were assessed by the Centre for Marine and Petroleum Technology and they said that the work made a “very significant contribution to the field”. It is my view that “As statisticians we don’t just crunch numbers, we also help to improve people’s lives”.

FAQs

Question: Once you had collected the data, how did you analyse it?
Answer: Using operational research techniques and statistical methods. These included: forecasting and time series analysis, Fourier analysis, extreme value analysis, and directional statistics.

Question: What software did you use?
Answer: It was completed using a combination of PASCAL, FORTRAN, SPlus and MATLAB.



References

The following articles provide relevant technical material, co-written with other academics (but they do not specifically relate to the story above):
1.     Weathering the storm - how OR steered a course between extreme statistics and offshore design (by John Bowers, Ian Morton and Gill Mould)
2.     Extreme value analysis in a multivariate offshore environment (by I.D. Morton, J. Bowers)
3.     Directional Statistics of the wind and waves (by J.A. Bowers, I.D. Morton, G.I. Mould)



[1] It’s nearly 60 metres from the helideck to the level of the sea – to put that in perspective that’s about the same as the width of a football pitch – a long way down.
[2] Swell waves – these are generated from distant storms, they have lost a lot of their energy. They are smooth waves and are rounded in structure.
[3] Wind sea – these are local storm waves with lots of spray. These waves are rough, unlike the smooth swell.
[4] If the wind speed is 40 metres per second, it’s a hurricane.

Saturday 17 November 2012

Confessions of an ageing socmedite (that's short for social-media-ite)


My aim is to help to improve peoples lives through the work I do as a statistician. To achieve this it seems sensible to publicise my statistical work. Someone recently commented that I should advertise my past in a blog – they said that it’s a bit like writing a CV, but with an added twist. Further to this, they added that “you should create a blog about your past, share case studies of your work, and [most importantly] receive feedback on these blogs”. That latter bit is the “added twist”, and it might be the part that most of us would admit to being frightened of. He went on to say that “by continually learning from the feedback, and becoming better known, you will be able to combine business knowledge and statistical methods to create good case studies of your work”.

And I initially thought, “simple, what a great idea”.

I confess that “getting into” blogging, and other social media tools, probably makes me a bit of a geek. Or in Seth Godin’s words, I have become one of the nerds that’s increasing in number.

But, before you think that I have flipped, you’ll be pleased to know that I won’t be racing headfirst into writing my blog – that’s because I am slowly realising how difficult it is to write about my work as a statistician. Yes, maybe my age is slowing me down a bit, or maybe it’s something to do with not being able to write a story. But I don’t think so. Also, the problem is nothing to do with numbers, or formulae, or this-and-that fancy analysis technique that statisticians use. It’s more to do with getting around those confidentiality agreements, and it’s also a bit about not wanting to cause any conflict with employers and/or colleagues – after all they might need me, or I might need them, in the future.

However, I will not be put off, and the seeds of this blogging challenge have been planted. So watch this space.

By the way, I recently read a great post on the Understanding Uncertainty website (Obtained from here http://understandinguncertainty.org/continuing-tragedy-l%E2%80%99aquila). It reminded me to (a) realise that I am learning all the time, and (b) that I do need to remember to keep the following points in the back of my mind:
  1. Think about uncertainty,
  2. Don’t give scientific advice if you aren’t scientifically qualified, and
  3. Don’t engage in informal communication about something you don’t know enough about.

Oh, and just in case you have missed it. I heard a great talk on TED and thought you might want to hear some ideas worth spreading. I like this one at the moment:





Follow me on Twitter: https://twitter.com/IDMorton001

Connect with me on LinkedIn: www.linkedin.com/in/idmorton

See my Presentations on SlideShare: http://www.slideshare.net/IDMorton001
Here is an example of a presentation I recently gave to a group of Primary School children: http://www.slideshare.net/IDMorton001/presentation-to-primary-school-government-statistician-13th-nov-2012

Please let me have your feedback.

Tuesday 25 September 2012

“Big traditional banks worst on customer satisfaction” or should that be “Comparing apples and pears” ?

It was interesting to see the production of a league table of Banks in the latest Which? magazine which ran with the headline “High-street banks failing on customer satisfaction: Consumer views of the big banks remains low”.

But as an ex-bank employee, and analyst/statistician this got me thinking …

Is customer satisfaction really that poor for these “Big high-street banks” ?

Is the position of First Direct, Santander, or Saga misleading ?

Is the league table spreading doom and gloom that is unjustified ?

Or, is customer satisfaction even worse for some banks than that shown in the league table ?

Don’t get me wrong, I am not defending any of these banks and I am not knocking Which? for highlighting the issue (I have had experience of poor customer service). I just want to say that the picture isn’t as straight forward as the league table suggests and I will (hopefully) explain why.

Here goes … Any particular bank in the league table will offer different products, and have different customers, to any other bank in the list – after all, they need to do this to suit their customers and to gain competitive advantage through niche products. But I believe this product offering, to different customers, has the impact that the age and gender make-up of customers at some banks will be different to the age and gender make-up of customers at other banks. THEY ARE LIKE APPLES AND PEARS.

Statisticians refer to the different make up of customers as “case mix” and they would also say that it should be “controlled for”. For a real example, here is a quote “Saga [bank] offer a range of competitive finance products exclusively for the over 50s”. I wouldn’t have thought that this particular quote would be applicable for some of the other banks, they will possibly have fewer customers aged over 50.

And this is where things get even more complicated … ideally if you want to compare customer satisfaction per se, you have to remove the “case mix” – you “control for” any differences that are due to age, gender and other factors.

So what I am saying is that the Which? League table may be slightly misleading. If you really want to compare on customer satisfaction itself you should do it AFTER “CONTROLLING FOR” THE CASE MIX.

This could produce an unwanted result, and the table may look completely different to the one published (e.g. Santander may be in the middle of the table, or First Direct may be way off the top etc). But, at least you wouldn’t be comparing apples and pears, YOU WOULD BE ADDRESSING CUSTOMER SATISFACTION ITSELF.