Addressing the challenges to sustainable operation of gas turbines
Summary
TLDRIn this webinar, John Douglas and Laurie Brooking explore the challenges of maintaining sustainable operational performance in gas turbines. They emphasize the use of physics-based models in asset management to optimize turbine longevity and performance. Laurie discusses the behaviour of high temperature materials, including single crystal super alloys, plagued by degradation and stress factors like creep and fatigue. The presentation includes insights into Fraser Nash's growth and capabilities in engineering and consultancy, focusing on integrating digital tools with robust engineering principles. The evidence indicates that adopting these physics-based strategies can better inform operational decisions, potentially leading to more efficient and flexible management of gas turbine assets, suggesting extensions in operational life and improved reliability.
Takeaways
- 🔧 Use of physics-based models and digital assets to manage gas turbine operations.
- 🌍 Fraser Nash's international growth, with operations in the UK and Australia.
- 📈 Importance of physics models in extending gas turbine operational life.
- ⚙️ Corrosion fatigue in high temperature materials like single crystal super alloys.
- 🔍 Emphasis on research collaboration for developing engineering techniques.
- 💡 Opportunities presented by flexible operation and maintenance deferral.
- 🔬 Detailed technical methods to assess and optimize turbine parts.
- 📊 Insights into understanding corrosion fatigue interactions.
- 📚 Strong academic ties enhancing research and development.
- 🚀 Fraser Nash's diversified engineering services across industries like aerospace and industrial applications.
Timeline
- 00:00:00 - 00:05:00
John Douglas welcomes participants to a webinar about sustainable operational gas turbines, explaining the broad subject will focus on physics-based models for asset management. Laurie Brooking introduces himself with a relevant background in high-temperature materials and research on gas turbine blade materials.
- 00:05:00 - 00:10:00
John shares his professional journey starting with a PhD in gas turbine materials and his current role at Fraser Nash overseeing accounts related to turbines. The presentation will highlight Fraser Nash's capabilities, specifically the Rotating Machinery Center and the use of physics-based models for digital asset management.
- 00:10:00 - 00:15:00
Fraser Nash has grown significantly, now with offices in the UK and Australia, employing around 900 staff. They focus on broad capabilities and collaboration with academic institutions to solve complex problems for clients often returning due to their comprehensive approach to challenging issues.
- 00:15:00 - 00:20:00
The Rotating Machinery Center at Fraser Nash specializes in design, digital assets, and engineering services. They cover a spectrum from gas turbines to other rotating machinery. Their approach emphasizes developing robust, physics-based digital asset management systems rather than just data analytics.
- 00:20:00 - 00:25:00
Using physics-based methods, Fraser Nash believes they get better results in managing complex machinery. Their hybrid approach integrates data with physics to extend machine operating life, boost performance, and optimize part placement. They often help clients defer maintenance based on comprehensive assessments.
- 00:25:00 - 00:30:00
The discussion delves into the technical methodologies, where simple models are initially used and refined based on emerging data. This is framed within a probabilistic approach to manage uncertainties from sensor data and model accuracy, ultimately supporting risk-based maintenance decisions.
- 00:30:00 - 00:35:00
Highlighting examples of engineering services, they have developed efficient ventilation standards for gas turbine enclosures, vital for health and safety, especially considering new standards like ATEX. Investigations into hydrogen's effects are ongoing.
- 00:35:00 - 00:40:00
Asset management plays a crucial role. The approach allows understanding of machine's damage status in real time, guided by physics-based assessments. This is valuable for operational decisions, data interpretation, and risk assessments under varying load conditions.
- 00:40:00 - 00:45:00
Focus shifts to Laurie Brooking's analysis of corrosion-fatigue interactions in single crystal alloys, showing propagation patterns in high-temperature environments. This segment underscores the complexity of predicting material behavior and the importance of understanding these interactions for reliability.
- 00:45:00 - 00:50:00
Tests reveal how corrosion fatigue emerges through intricate mechanisms. Laurie observes how different elements and deposits influence crack initiation and propagation, indicating these materials' response varies significantly with conditions impacting a turbine's component lifespan.
- 00:50:00 - 00:55:00
Laurie presents how physics models like Paris curves are adapted to monitor crack growth in fatigue tests. Different rates of corrosion-fatigue interactions are mapped out, showing variance in material responses and the benefits of incorporating this understanding into turbine operation.
- 00:55:00 - 01:03:16
Q&A addresses standstill corrosion and uncertainty in sensor data, emphasizing physics models' adaptiveness and reliability compared to pure data analytics. The session closes with John acknowledging Rizvan's contributions and promising further discussions on probabilistics and risk management.
Mind Map
Video Q&A
Who are the main speakers of the webinar?
John Douglas and Laurie Brooking.
What company do the speakers represent?
Fraser Nash.
What is the main focus of the webinar?
Addressing the challenges to sustainable operational gas turbines using physics-based models.
What is Laurie Brooking's area of expertise?
High temperature materials and gas turbine blade materials, with a focus on degradation and super alloys.
What company is mentioned as having a growing international presence?
Fraser Nash, with offices in the UK and Australia.
What are some applications of the discussed physics-based models?
Applications include asset management and optimizing the operational integrity of gas turbines.
How does physics-based modeling benefit gas turbines?
It helps manage damage accumulation, optimize operations, and extend the life of turbine components.
Can the methods discussed help with part-load operation of gas turbines?
Yes, they can optimize and extend part-load operation conditions.
View more video summaries
The Man With The Seven Second Memory
Data Communication - N10-008 CompTIA Network+ : 1.1
"If You Ever Freeze Time, Make Sure To Follow The Rules" | Creepypasta | Scary Story
How to Collect a Herbarium Specimen and Tissue Sample
Thales Teixeira Introduces "Decoupling" as the Customer-centric Approach to Digital Disruption
Ryan Reynolds Is The New Steve Jobs
- 00:00:08ah
- 00:00:09good morning good afternoon everybody um
- 00:00:12right on cue my builder has just got his
- 00:00:13tools out and started creating noise i
- 00:00:15hope you can't hear that too badly
- 00:00:17um it's john douglas i'm just um i'm
- 00:00:19just going to give it another minute or
- 00:00:21so
- 00:00:21to to let some more people
- 00:00:24uh sign into this and then we'll get
- 00:00:28started
- 00:01:42okay so good afternoon everybody um i
- 00:01:45think most of the people who are going
- 00:01:46to join are probably joined already now
- 00:01:48so
- 00:01:49i'm going to get started um thank you to
- 00:01:52everyone for
- 00:01:53for taking an hour out of their time to
- 00:01:55to listen to this webinar
- 00:01:57on addressing the challenges to
- 00:01:59sustainable operational gas turbines
- 00:02:01my name is john douglas i've got an
- 00:02:03introduction slide coming up
- 00:02:05i'm going to talk through obviously this
- 00:02:06is a very big subject area um
- 00:02:08i'll talk briefly about some of the work
- 00:02:11that we at phrase and ash have been
- 00:02:12doing
- 00:02:13in this very big subject area and i
- 00:02:16obviously can't cover all of what we're
- 00:02:17doing and all of what happens in this
- 00:02:19area but hopefully
- 00:02:20hopefully you're going to find what's
- 00:02:21what's presented today of interest
- 00:02:24um so
- 00:02:28next slide um hopefully laurie you can
- 00:02:30see that
- 00:02:31i don't know if you want to give us a
- 00:02:32brief introduction to yourself before we
- 00:02:34get
- 00:02:34started yeah thanks john yeah so
- 00:02:37i'm laurie brooking and um yeah my
- 00:02:40interest in high temperature materials
- 00:02:42sort of started with my phd at
- 00:02:44cranfield university where i was looking
- 00:02:46primarily at
- 00:02:47degradation and gas turbine materials
- 00:02:50gas turbine blade materials so single
- 00:02:51crystal
- 00:02:52uh super alloys and super alloys um and
- 00:02:55and after kind of being in in the
- 00:02:57research environment there i then moved
- 00:02:59over to
- 00:03:00consultancy and and started at fraser
- 00:03:02nation
- 00:03:03sort of since since then have been
- 00:03:05continuing to work in in largely high
- 00:03:07temperature materials and after
- 00:03:08integrity so
- 00:03:09so looking at how how we can help our
- 00:03:11clients to
- 00:03:12improve the integrity of their assets
- 00:03:14and longevity and
- 00:03:16get the best cost benefits so yeah
- 00:03:18that's me
- 00:03:20thanks very much laurie um so i guess my
- 00:03:23my interest started very similarly to
- 00:03:25lorries with with a phd
- 00:03:27in uh in gaster turbine materials as
- 00:03:29well i'm a mechanical engineer but i was
- 00:03:31looking at
- 00:03:31single crystal alloys from a creep
- 00:03:34fatigue perspective rather than from a
- 00:03:36degradation
- 00:03:36oxidation type point of view and over
- 00:03:39the years i've been with fraser nash for
- 00:03:42over 17 years now in various roles more
- 00:03:45recently
- 00:03:47on the commercial side looking after the
- 00:03:50the account for
- 00:03:51um turbines and i've recently set up the
- 00:03:55rotating machinery center which which
- 00:03:56i'll talk a little bit more about
- 00:03:59as we go through the slides um
- 00:04:02in terms of the presentation today
- 00:04:05hopefully the content slide has come up
- 00:04:07um there's opportunity obviously to
- 00:04:09advertise fraser nash so a quick slide
- 00:04:11on
- 00:04:12praising ash without making it too much
- 00:04:14of an advertisement but really just a
- 00:04:15bit
- 00:04:16brief of a brief on who we are and
- 00:04:19um what we look like today moving on
- 00:04:22from that
- 00:04:23my little corner of fraser nash which is
- 00:04:24the rotating machinery center
- 00:04:26and some of the work that we're doing
- 00:04:28within within
- 00:04:29within within within that capability
- 00:04:31area um
- 00:04:34and really the presentation today um is
- 00:04:37is is about the use of physics-based
- 00:04:39models for
- 00:04:40asset management for one of a better
- 00:04:42word digital is one of those words
- 00:04:44that's used quite a lot at the moment
- 00:04:46what we're doing does fit within the
- 00:04:48description of digital
- 00:04:49uh one way or another so we we call it
- 00:04:52digital assets
- 00:04:54there's a lot of talk out there about
- 00:04:55data and analytics um
- 00:04:57and there is a place for that but with
- 00:04:59complex machinery like this
- 00:05:01we find that the use of physics-based
- 00:05:03methods or mechanical engineering
- 00:05:05uh gives us a better result in some in
- 00:05:08some instances
- 00:05:09and so there's various parts of that
- 00:05:11physics uh i'll get into
- 00:05:12some of the detail on that but really um
- 00:05:16once i've given an overview of that
- 00:05:17approach it's over to laurie to
- 00:05:19to to a presentation on on what is what
- 00:05:22is what he's looking at in terms of
- 00:05:24corrosion fatigue interactions
- 00:05:26you can see with crystal alloys um a
- 00:05:28very particular corner of those physics
- 00:05:30models
- 00:05:32the next slide oops i think i've gone
- 00:05:36too
- 00:05:37so a little bit about frasier nash i've
- 00:05:40been here for 17 years we were less than
- 00:05:42100 people when i joined
- 00:05:43we've had quite a growth trajectory uh
- 00:05:45we've moved out of the uk and we're now
- 00:05:47in australia as well so
- 00:05:49we have a significant number of offices
- 00:05:50in the uk we've got four officers in
- 00:05:52australia
- 00:05:53um i hope i'm not misrepresenting the
- 00:05:55australia officers
- 00:05:57um in terms of they've been there around
- 00:06:00about 10 years
- 00:06:01and all right there's over 100 people in
- 00:06:03those offices now or if it's getting
- 00:06:05closer to 200
- 00:06:06people but it's that sort of order of
- 00:06:07magnitude in total we've got 900
- 00:06:10ish permanent staff in our offices in
- 00:06:13the uk
- 00:06:14and in australia we have a number of
- 00:06:16other people positioned around the globe
- 00:06:17in austria
- 00:06:18in in america and europe on the site
- 00:06:20with clients doing some interesting work
- 00:06:22as well
- 00:06:23um if we haven't got an expert within
- 00:06:26the company that can help
- 00:06:27then we have access to a network of over
- 00:06:30three thousand other
- 00:06:31host associates who we can call on for
- 00:06:33expertise
- 00:06:34um so we have quite a broad network
- 00:06:37very very broad capability across
- 00:06:40three to four thousand people and if we
- 00:06:43still can't find someone that can help
- 00:06:45with a problem
- 00:06:46then we have very strong collaborations
- 00:06:48with with the universities
- 00:06:50the vast majority well a large
- 00:06:52proportion of the people at frozen ash
- 00:06:54have done some important post-doctoral
- 00:06:56study or
- 00:06:56sorry postgraduate study either masters
- 00:06:59phds
- 00:07:01postdoctoral and so very strong links to
- 00:07:03academia
- 00:07:04and and if and if we can't solve it if
- 00:07:07our associates can't solve it then we
- 00:07:09can call on
- 00:07:10those collaborations to to get experts
- 00:07:12in to help with whatever problems
- 00:07:14um we're also getting something right a
- 00:07:16lot of our customers keep coming back
- 00:07:18um for for more and more help um
- 00:07:21and that's great there's various reasons
- 00:07:23for that but i think i think the biggest
- 00:07:25reason for that
- 00:07:25is the fact that um we don't just we
- 00:07:27don't just put people on site
- 00:07:29and sell people for man hours if you
- 00:07:31like we
- 00:07:32we like to own our customers problems
- 00:07:35and we like to provide solutions to
- 00:07:37those problems
- 00:07:37and i think it's that it's that approach
- 00:07:40which gives us an
- 00:07:41edge in the marketplace so
- 00:07:44that's that's i guess that's the
- 00:07:46advertisement over for fraser nash
- 00:07:47um if you want to find out more you can
- 00:07:49call me or you can get in touch by the
- 00:07:51website
- 00:07:52we might be able to help you moving on
- 00:07:55um my little corner of frasier nash is
- 00:07:58is is the rotating machinery center
- 00:08:00uh recently set up just before christmas
- 00:08:03um and we've launched a
- 00:08:04a part of the website which is fnc.co uk
- 00:08:08forward slash rmse and
- 00:08:11broadly speak about that that gives you
- 00:08:13a brief overview of
- 00:08:14some of the skill sets that we've got in
- 00:08:16in the rotating machinery
- 00:08:18field uh the vast majority of it is
- 00:08:20around gas turbines
- 00:08:21steam turbines pumps that type of stuff
- 00:08:23but actually we do a lot of work on on
- 00:08:25other rotating machinery
- 00:08:27including energy storage devices and uh
- 00:08:30rockets and various other gadgets so
- 00:08:33we've got we've got a vast
- 00:08:34experience uh we even do stuff on you
- 00:08:37know
- 00:08:37fans for kitchen appliances and that
- 00:08:39type of stuff so so
- 00:08:42we've got a broad experience there and
- 00:08:43if you go to the website you'll find out
- 00:08:46uh broadly speaking we've got three
- 00:08:47service areas
- 00:08:49machinery design a lot of our a lot of
- 00:08:51our engineers
- 00:08:52have had careers at original equipment
- 00:08:56manufacturing companies and brought a
- 00:08:59lot of that design experience with them
- 00:09:01others have spent time on site with us
- 00:09:03with those customers and
- 00:09:05and gained a lot of experience in terms
- 00:09:06of design and so we
- 00:09:08you know whether it's a component design
- 00:09:10for a blade or a disc or a
- 00:09:12combustor or whether it's a full system
- 00:09:15or a stage or a turbine we can pick up
- 00:09:18that kind of work
- 00:09:19and depending on the type of the type of
- 00:09:21machine that we're looking at we can
- 00:09:22actually design a full machine
- 00:09:24and give you a set of drawings and we
- 00:09:25can work with you to get that
- 00:09:26manufactured so
- 00:09:27we have a significant significant
- 00:09:29machinery design
- 00:09:30capability there and and some of that is
- 00:09:33is outlined on the website um
- 00:09:36the the next part of our service
- 00:09:38capability there is digital assets and
- 00:09:40that's what we're going to be talking
- 00:09:41about today so i won't go into too much
- 00:09:42detail here
- 00:09:43um and then engineering services if you
- 00:09:46know we've got a broad range of services
- 00:09:48we've got a lot of engineers doing a lot
- 00:09:49of different things
- 00:09:50in a lot of different teams um if it's
- 00:09:52not covered by
- 00:09:53machinery design or digital assets then
- 00:09:55there are a list of services there
- 00:09:57and there's a brief list here which
- 00:09:59covers thermodynamic performance
- 00:10:01road dynamics materials that type stuff
- 00:10:05and before we get into the talk on
- 00:10:06digital assets then
- 00:10:08and then into laurie's presentation just
- 00:10:10a quick example
- 00:10:11of a a service area
- 00:10:14under the engineering services so
- 00:10:17hopefully the sliders refreshed
- 00:10:19there's a picture there of a gas cloud
- 00:10:21within a within a
- 00:10:22gas turbine enclosure our team of fluid
- 00:10:25engineers have been working with
- 00:10:27customers been working with health and
- 00:10:30safety laboratories to develop
- 00:10:32standards around atex and dasia
- 00:10:35in terms of ventilation of enclosures
- 00:10:38and
- 00:10:39a lot of that work over the last 20
- 00:10:40years has been focused primarily on
- 00:10:42methane gas clouds
- 00:10:44where do they form how do they form are
- 00:10:46they ventilated and if they're not
- 00:10:48ventilated where do we need to put
- 00:10:49sensors
- 00:10:50in that type of work like so the team
- 00:10:52have been working on this for a very
- 00:10:53long time
- 00:10:54they've developed some very efficient
- 00:10:55fluid dynamics methods which which cover
- 00:10:58which cover the basics of how to how to
- 00:11:00get that done
- 00:11:02and how to get that done really quickly
- 00:11:03really efficiently so if you have that
- 00:11:05kind of problem you really should be
- 00:11:06talking to our team
- 00:11:08uh they're central to the development of
- 00:11:09these standards and understand how to
- 00:11:11solve these problems
- 00:11:12and understand how to get you certified
- 00:11:14in those in those sorts of problems
- 00:11:17more recently we've been that team have
- 00:11:19been working on
- 00:11:20um how those standards can be modified
- 00:11:23how they're affected by the introduction
- 00:11:25of hydrogen
- 00:11:26and whether that's hydrogen at one
- 00:11:28percent or one hundred percent and
- 00:11:30anything in between
- 00:11:31uh those simulations are underway and we
- 00:11:33are working with
- 00:11:34we are working with uh various partners
- 00:11:36to develop that standard and to actually
- 00:11:38look at
- 00:11:39uh installations within europe um
- 00:11:42european facilities at the moment so
- 00:11:46it's a little bit it's a little bit left
- 00:11:48field but that's that's just one of the
- 00:11:50services that we can offer
- 00:11:51um it's worth having a look at the
- 00:11:52website to see which other services
- 00:11:55we offer it's still under development so
- 00:11:57not everything that we offer is on there
- 00:11:59and if you want to find out more again
- 00:12:00please call me
- 00:12:01or get in touch by the website
- 00:12:05getting into what are we doing on the
- 00:12:08on on the asset side um i guess a lot of
- 00:12:11people on the call would be aware
- 00:12:13that the the you know the oems um
- 00:12:16power stations um owners operators
- 00:12:20um have got a lot of telemetry on on gas
- 00:12:23turbines tomorrow that's a
- 00:12:24small unit pumping gas along an oil and
- 00:12:27gas pipeline or whether that's a
- 00:12:29large power station there's for a long
- 00:12:31time a lot of data has been coming off
- 00:12:33there for
- 00:12:34the purpose of vibration monitoring
- 00:12:35performance monitoring
- 00:12:38we've figured out over the last five or
- 00:12:40ten years how to make best use of that
- 00:12:42data
- 00:12:43in terms of understanding what's
- 00:12:44happening with the machine and
- 00:12:47even just at a simple level am i running
- 00:12:50a full load steady state or am i
- 00:12:52running at part load 50 um
- 00:12:55we've got some real-time algorithms
- 00:12:57which look at the data
- 00:12:58understand what level the machine's
- 00:13:01operating at
- 00:13:02and then we've got lots of physics
- 00:13:04models we've got lots of algorithms
- 00:13:06which which will get into the detail of
- 00:13:08what's happening with metal temperature
- 00:13:11and metal stress within the components
- 00:13:13that are of interest
- 00:13:14um so if you're running at full load
- 00:13:16steady state you're expecting various
- 00:13:18failure modes within blades or discs or
- 00:13:20nozzles
- 00:13:21uh we're monitoring the stresses we're
- 00:13:23moderating monitoring
- 00:13:25the temperatures in those components
- 00:13:27real time
- 00:13:28and we're able to understand from that
- 00:13:30how much damage is accumulating
- 00:13:32within within within those components
- 00:13:35and
- 00:13:35you know you can integrate that
- 00:13:36throughout the life of that that turbine
- 00:13:41we do that with various physics based
- 00:13:42approaches including creep
- 00:13:44fatigue corrosion and oxidation my
- 00:13:47interests are around greek fatigue
- 00:13:49lori is very interested in the corrosion
- 00:13:51and oxidation end of things
- 00:13:52but we bring all of those models
- 00:13:53together to understand the condition of
- 00:13:56your machine
- 00:13:57at any given point in time and you know
- 00:13:59there's various advantages or various
- 00:14:01opportunities
- 00:14:02coming out of that um not least flexible
- 00:14:04operation so
- 00:14:06can i can i run my machine at 50 power
- 00:14:09for
- 00:14:09for longer than the usual maintenance
- 00:14:11window the answer is yes you can
- 00:14:12generally speaking
- 00:14:14um part load power is generally um
- 00:14:18less damaging than full well steady
- 00:14:20state or than the design case
- 00:14:21um but not always and that's where the
- 00:14:23physics comes in so data analytics on
- 00:14:26its own
- 00:14:27um can't tell you when you're going to
- 00:14:29bake the back end of the gas turbine for
- 00:14:30instance
- 00:14:31um and so so running a part load
- 00:14:33generally it's less damaging but there
- 00:14:35are instances where
- 00:14:36you will cause more damage by running at
- 00:14:38reduced load and it's the physics models
- 00:14:40that really get into the nitty-gritty of
- 00:14:41that
- 00:14:41that tell you to tell you why and and
- 00:14:44when
- 00:14:45those things are happening um so
- 00:14:47flexible operation
- 00:14:48can i extend the life well the physics
- 00:14:50will tell me whether i can or not
- 00:14:53what if what if i want to run a power
- 00:14:55boost what if i want to do 120
- 00:14:57for a couple of hours a month what does
- 00:14:59that do to my maintenance window what
- 00:15:00does that do to the damage within the
- 00:15:02engine
- 00:15:02and that's that's where that's where the
- 00:15:04value of these sorts of
- 00:15:05algorithms come in um can i optimize
- 00:15:09part placement to recover pop
- 00:15:11performance of
- 00:15:12the gas turbine yeah you can so we can
- 00:15:14monitor the engine we can use the
- 00:15:16physics
- 00:15:16to understand when you might want to do
- 00:15:19a wash on the on the on the on the
- 00:15:20compressor
- 00:15:21or when you might need to replace some
- 00:15:23nozzles which have which will start to
- 00:15:25build up uh product on them
- 00:15:27uh or or or or durability issues um
- 00:15:30you know so when to replace parts when's
- 00:15:34a good time to replace parts with
- 00:15:36that sort of stuff falls out of physics
- 00:15:37and then um
- 00:15:39maintenance deferring um i know my
- 00:15:41engine
- 00:15:42is coming up for maintenance within the
- 00:15:44next three months but actually
- 00:15:46i want to run it for six months what was
- 00:15:48my previous what was my previous
- 00:15:50use case done to to the the condition of
- 00:15:53those parts and can i can i defer my
- 00:15:55maintenance while i get through this
- 00:15:56busy period for the next six months
- 00:15:58and all sorts of questions and that's
- 00:16:00and that's the kind of stuff that we're
- 00:16:01working with power stations
- 00:16:02working with oems to to solve and using
- 00:16:05these
- 00:16:06methods to to actually do so a little
- 00:16:09bit
- 00:16:10a little bit more detail on on without
- 00:16:13getting into the physics a little bit
- 00:16:15more detail on the technical approach
- 00:16:17um we start simple and and we build from
- 00:16:20there so
- 00:16:21we've got some very very simple models
- 00:16:24that
- 00:16:24that that tell you about what's going on
- 00:16:27in the physics of crete fatigue
- 00:16:29corrosion oxidation there's no point in
- 00:16:32us developing
- 00:16:33highly complex models for your engine uh
- 00:16:37on all on all corners without the best
- 00:16:39the best approaches to is to start
- 00:16:41with simple models and then start
- 00:16:42looking at the results that are coming
- 00:16:44off
- 00:16:44and we'll find out very quickly whether
- 00:16:46your engine engines
- 00:16:48are a creep dominated fatigue dominated
- 00:16:51or if they've got oxidation and
- 00:16:52corrosion problems
- 00:16:54um and from that we can then start to
- 00:16:57build in
- 00:16:58more and more detail on on the physics
- 00:17:00and uh and
- 00:17:02and uh and and take it from there um
- 00:17:04there's no point in building a highly
- 00:17:06complex fatigue model if you've got a
- 00:17:08predominate machine
- 00:17:09um that's the first point i guess the
- 00:17:12next point is that
- 00:17:13this is all about risk um we
- 00:17:18we obviously need to understand what's
- 00:17:19going on with the creep and the fatigue
- 00:17:20for various reasons but actually
- 00:17:24the sensors the data that's coming up
- 00:17:26with sensors isn't 100 accurate
- 00:17:28what's going on with thermocouple drift
- 00:17:30is that thermocouple accurate
- 00:17:32is it even working those sorts of
- 00:17:34questions come into your risk based
- 00:17:35model
- 00:17:36what about my life and model my creep
- 00:17:38model if i'm starting with a basic creep
- 00:17:40model
- 00:17:40then then there's obviously a lot better
- 00:17:42certainty on the accuracy of that model
- 00:17:45versus versus a mob spell model which is
- 00:17:47being developed
- 00:17:48very very neatly for your application
- 00:17:50and so
- 00:17:51understanding the uncertainty
- 00:17:53understanding
- 00:17:54the risk that comes with that
- 00:17:56uncertainty is is a big part of what
- 00:17:58we're doing
- 00:17:59and all of what we're doing here sits
- 00:18:00within a probabilistic wrapper
- 00:18:04those methods have been developed from
- 00:18:05monte carlo simulations of graphite
- 00:18:07cores on nuclear reactors so
- 00:18:09we're pretty confident that those
- 00:18:11methods work and we're pretty confident
- 00:18:13with them with the answers that come out
- 00:18:14of that
- 00:18:16um and i guess the the last point on
- 00:18:18this slide really is about the the fact
- 00:18:20that we're not
- 00:18:21you know we're not hungry for more and
- 00:18:23more data um
- 00:18:24i mean it's obviously always quite nice
- 00:18:26to have more data but
- 00:18:28the more data you have the bigger the
- 00:18:29problem you've got processing it um
- 00:18:31there's a lot of machinery out there
- 00:18:32with with with
- 00:18:34with sensors what we try and do is make
- 00:18:38best use of that sensor data
- 00:18:40without actually needing to go in and
- 00:18:42refit new sensors
- 00:18:43and and new telemetry and just try and
- 00:18:46make use of what we've got
- 00:18:47occasionally we're coming to a situation
- 00:18:49where we just don't have enough data
- 00:18:51and it is sensible to to put more
- 00:18:53sensors where we need them
- 00:18:55but most of the time we we get away with
- 00:18:59with just the sense of data that exists
- 00:19:03so it's not a big expensive job to refit
- 00:19:06your fleet
- 00:19:09and then just turning some of that into
- 00:19:11it into a picture quickly
- 00:19:12um on the left-hand side we've got the
- 00:19:15data platform
- 00:19:17on the right-hand side we've got a
- 00:19:18risk-based decision making tool
- 00:19:21uh and in between we've got our
- 00:19:22physics-based tool set
- 00:19:24um just just to you know on the first
- 00:19:27level we've got some very simple models
- 00:19:29um we will implement them very quickly
- 00:19:32very cheaply to start producing results
- 00:19:35and
- 00:19:35and once once we start to get results
- 00:19:37from that kind of approach that's when
- 00:19:39we can start to think about do we need a
- 00:19:41creep fatigue interaction model do we
- 00:19:43need a more complex corrosion
- 00:19:45or oxidation model uh and from that
- 00:19:48do we then need to get into the various
- 00:19:50specialist models that
- 00:19:51that can be developed on a needs must
- 00:19:55basis
- 00:19:57hopefully hopefully that's given you
- 00:19:59quite a nice overview of our
- 00:20:01approach to managing your assets using
- 00:20:04using sort of digital
- 00:20:08techniques coupled with with
- 00:20:10physics-based approaches
- 00:20:14i can't talk in detail about all of that
- 00:20:16i'm not going to talk in detail
- 00:20:18about all of that i think at this stage
- 00:20:20really that's a little bit
- 00:20:21that's that's enough of an understanding
- 00:20:23to understand our approach
- 00:20:25um i think i'm just going to hand over
- 00:20:27to lori now so that he can talk a little
- 00:20:29bit about his corrosion fatigue
- 00:20:30interactions
- 00:20:31and what he's seeing on on single
- 00:20:33crystal alloys
- 00:20:38great so um hello everyone um yeah my
- 00:20:41name is
- 00:20:41laurie brooking and uh i'll be taking a
- 00:20:44quick look at corrosion fatigue
- 00:20:46interactions in
- 00:20:47single crystal super alloys
- 00:20:51so this is a common issue and it's seen
- 00:20:53both in industrial and aerospace
- 00:20:56gas turbines and it's
- 00:20:59a cracking morphology often similar to
- 00:21:02fatigue
- 00:21:03and that's observed typically in hot gas
- 00:21:05pump components
- 00:21:06you know earlier and sooner than
- 00:21:08predicted so we'll be talking about what
- 00:21:10causes this issue
- 00:21:12how do these cracks propagate and how we
- 00:21:14can look to predict
- 00:21:16and mitigate against against this
- 00:21:18problem so on
- 00:21:20um just just just at the right here of
- 00:21:23the bullet points you can see there's a
- 00:21:24couple of images of uh micrographs
- 00:21:27uh cmsx4 single crystal super if you're
- 00:21:29not familiar so i was just going to kind
- 00:21:31of explain what is a single crystal c
- 00:21:32probably
- 00:21:33why they're different to to a lot of
- 00:21:35other allies so you can see we've got
- 00:21:37that
- 00:21:38cuboid cuboidal type micro structures
- 00:21:40camera gamma prime microstructures for
- 00:21:4231
- 00:21:43and the gamma prime is is the kind of
- 00:21:45the cuboidal feature the square feature
- 00:21:47there
- 00:21:49and the matrix is the bit in between
- 00:21:52those cuboidal features
- 00:21:53so they're both face centered cubic um
- 00:21:56and
- 00:21:56the gamma prime is an l one two ordered
- 00:21:59face
- 00:21:59centered cubic and just next to that the
- 00:22:01really high mag
- 00:22:02five nanometer image is showing you the
- 00:22:05interface between that camera and the
- 00:22:06camera prime so really this is
- 00:22:08this is what makes these materials a bit
- 00:22:09special so whilst you've got two phases
- 00:22:12because the unit cell size is so
- 00:22:14comparable uh we actually
- 00:22:16don't really see a phase boundary you
- 00:22:18can see that the lattice coherence is
- 00:22:20really very good
- 00:22:21and in in these materials certainly at
- 00:22:24this temperature obviously things change
- 00:22:25around a bit with different temperatures
- 00:22:26but
- 00:22:26we don't really have a grain boundary
- 00:22:28hence why we call it a single crystal
- 00:22:30super light and that comes with numerous
- 00:22:32advantages for high temperature
- 00:22:34material properties creep fatigue things
- 00:22:37like that
- 00:22:38along the bottom i've got some images of
- 00:22:40what typical corrosion fatigue damage
- 00:22:42looks like so what we're talking about
- 00:22:43in this presentation the type of damages
- 00:22:45we we
- 00:22:46would see so the image on the bottom
- 00:22:48left there
- 00:22:49is of an industrial gas turbine and we
- 00:22:51can see
- 00:22:52a corrosion damage on the platform where
- 00:22:54the arrows are pointing to and
- 00:22:56actually bits of material has broken off
- 00:22:58there the middle image
- 00:23:00is undecration platform and again this
- 00:23:02is quite typical of what we would see
- 00:23:04um so it's a highly stressed area of the
- 00:23:06blade but actually lower temperatures
- 00:23:08than a lot of other other parts of the
- 00:23:09blade or some other parts of the blade
- 00:23:11but we see colonies of cracks
- 00:23:13propagating through the platform outside
- 00:23:15through
- 00:23:16through the root region that highly
- 00:23:18stressed region there
- 00:23:19and then the image on the bottom right
- 00:23:21is actually an image of
- 00:23:22what can happen if if this goes
- 00:23:24undetected you know obviously this
- 00:23:26happens
- 00:23:27a lot sooner other sort of said than
- 00:23:29expected in many cases
- 00:23:30and if it's not noticed the image on the
- 00:23:33right there is is some blades which have
- 00:23:34been ejected at the back of
- 00:23:36an aircraft engine actually upon takeoff
- 00:23:38so certainly not
- 00:23:39something that you want to be happening
- 00:23:41and
- 00:23:46a bit of background then into why this
- 00:23:49is happening
- 00:23:50and we believe the cracking occurs as a
- 00:23:52result of simultaneous deposit induced
- 00:23:54hot corrosion i'll talk a bit more about
- 00:23:56what that is uh in a bit combined with
- 00:23:59loading
- 00:23:59or stress and and the cracking can occur
- 00:24:03as the static loads as well as as as
- 00:24:06fatigue cycling so it's it's kind of a
- 00:24:08bit of a stress corrosion cracking type
- 00:24:10mechanism similar to what we would see
- 00:24:12you know in aqueous uh stress corrosion
- 00:24:15cracking at
- 00:24:15obviously much lower temperatures
- 00:24:17typically and i think a very
- 00:24:19very good question is why is this
- 00:24:20mechanism become prevalent in
- 00:24:22recent gas turbine designs and i think
- 00:24:25there's a couple of
- 00:24:26a couple of reasons for that the first
- 00:24:28is increases
- 00:24:29in turbine efficiencies you're driving
- 00:24:32up
- 00:24:33operating temperatures due to your car
- 00:24:35not cycle
- 00:24:36and we see the extended effect of hot
- 00:24:38corrosion so
- 00:24:39really we're pushing the boundaries and
- 00:24:41trying to improve efficiencies of these
- 00:24:43turbines or running some different fuels
- 00:24:46hydrogen etcetera which might need us to
- 00:24:48look at running at higher temperatures
- 00:24:49and
- 00:24:50and that can really extend the effects
- 00:24:52of hot corrosion into areas where we
- 00:24:54wouldn't typically see it
- 00:24:55and the second use is sort of related so
- 00:24:58because again we want to achieve higher
- 00:25:01temperatures to improve efficiencies
- 00:25:03we're looking at using materials um
- 00:25:05which have higher gamma prime factors
- 00:25:07factors fractions sorry due to
- 00:25:10um their you know sorry they've got high
- 00:25:13gamma prime fractions in order to
- 00:25:15achieve uh better high temperature
- 00:25:17properties so better creek properties
- 00:25:19and things like that
- 00:25:19but by increasing the gamma prime
- 00:25:21fraction we actually
- 00:25:23reduce uh the the refractory refractory
- 00:25:26element
- 00:25:26um concentration uh weight percentage
- 00:25:29and those refractory elements are really
- 00:25:31very good for building up protective
- 00:25:33oxides so
- 00:25:34we typically see with a lot of the
- 00:25:36really good high temperature super
- 00:25:37alloys
- 00:25:38they're a bit worse particularly at
- 00:25:39lower temperature corrosion and so
- 00:25:42there's a bit of a trade-off there and
- 00:25:43that's another reason i think we're
- 00:25:44seeing this mechanism become more
- 00:25:47more of an issue as we use those real
- 00:25:49high temperature
- 00:25:50high gamma prime fraction superb
- 00:25:56so i was just going to introduce high
- 00:25:59temperature deposit
- 00:26:00induced corrosion or hot corrosion
- 00:26:02because it's not necessarily that
- 00:26:04that well known about so it's typically
- 00:26:06defined by
- 00:26:07two mechanisms low temperature or type 2
- 00:26:10hot corrosion and high temperature or
- 00:26:12type 1 upgrade and you can see there
- 00:26:14on the figure on the right we've got a
- 00:26:16fairly linear relationship between
- 00:26:18temperature
- 00:26:19and oxidation rate however you have
- 00:26:21these two humps
- 00:26:22that accelerated damage um for type two
- 00:26:25and type one with type 2 occurring at
- 00:26:27slightly low
- 00:26:28low temperature and the reason we get
- 00:26:30these two hunts is
- 00:26:32largely due to chemistry so hot
- 00:26:34corrosion occurs as a result of
- 00:26:35corrosive species
- 00:26:37forming a liquid melt or eutectic on the
- 00:26:39surface of the blade
- 00:26:41and it typically um is is forming
- 00:26:44due to gaseous conditions and and also
- 00:26:47the deposition of
- 00:26:49of those species so depending on your
- 00:26:52your alloy
- 00:26:53composition your coating composition
- 00:26:55your deposit composition
- 00:26:57you can see the temperature of these
- 00:26:58these mechanisms move around
- 00:27:00and actually historically that that
- 00:27:02craft there on the right is
- 00:27:04is often thought of as as a kind of
- 00:27:06baseline but i think in reality
- 00:27:08the chemistry is quite a lot more
- 00:27:09complex and we see a bit more of a
- 00:27:11probably complex picture in in reality
- 00:27:15um so it's typically thought that um
- 00:27:18operation is electrochemical so that
- 00:27:21liquid deposit or liquid mixture forming
- 00:27:24on the blade is is a
- 00:27:25is a an ion transport and um
- 00:27:28is enabling uh acidic dissolution
- 00:27:32to happen kind of you know localized
- 00:27:34locations and cause
- 00:27:35cause this is cracking but just a
- 00:27:38general note
- 00:27:38um on on alloys and and how they tend to
- 00:27:41respond to this mechanism
- 00:27:43generally speaking chromia forming alloy
- 00:27:45solids which form a
- 00:27:46chromium oxide they're more resistant to
- 00:27:48type 2 and lower temperature oxidation
- 00:27:50and corrosion
- 00:27:51and your alumina forming alloys are more
- 00:27:53protective for type 1 and higher
- 00:27:54temperature
- 00:27:55oxidation and corrosion
- 00:27:59i also i think it's it's um
- 00:28:02sometimes debated a little bit as as to
- 00:28:05whether on or not hot corrosion
- 00:28:07reactions are deposit regulated
- 00:28:09or whether they're regulated by gaseous
- 00:28:11composition or
- 00:28:12partial pressure of sox or so3 and i
- 00:28:15suppose in my view
- 00:28:16again i think it's uh there's several
- 00:28:18computing mechanisms and
- 00:28:19it's it's quite a complex picture
- 00:28:21however i suppose
- 00:28:22that notes in industrial gas turbines um
- 00:28:25you often have
- 00:28:26uh filtration um which is limiting the
- 00:28:29amount of
- 00:28:30solid salt or deposit species or species
- 00:28:34which can be
- 00:28:35just through the engine of the positive
- 00:28:36so i think and a lot of the time we may
- 00:28:38see more of a
- 00:28:40kind of a factor of your gaseous in
- 00:28:43terms of
- 00:28:44limiting uh your hot corrosion um
- 00:28:47however in aerospace engines we don't we
- 00:28:48don't have that same filtration
- 00:28:50um so i think a lot of the time
- 00:28:53composition
- 00:28:53and deposit composition can play a
- 00:28:55bigger part
- 00:28:57in an aerospace application so again i
- 00:28:59think it can vary a little bit and
- 00:29:01there's there's quite a lot of complex
- 00:29:03chemistry going on
- 00:29:04as well
- 00:29:08so if we take a look then at some
- 00:29:09experimental test results here
- 00:29:11um these are a couple of images of
- 00:29:14statically loaded specimens so there's
- 00:29:15no cycling here these are cracks which
- 00:29:17have been generated
- 00:29:18uh you know in line with kind of stress
- 00:29:21corroding cracking rather
- 00:29:22rather than fatigue and what we see is
- 00:29:25multiple initiation sites
- 00:29:27and so the top image there is all the
- 00:29:30plain cylindrical
- 00:29:31specimen again statically loaded at
- 00:29:32quite high stretch there 911 pascals
- 00:29:35but we see uh multiple crack initiate
- 00:29:38cracks initiate and propagate um
- 00:29:41and uh obviously it's it's very much
- 00:29:45dependent on on the corrosion so if you
- 00:29:47take away that corrosion you obviously
- 00:29:48don't expect these materials to
- 00:29:50to initiate propagate cracks at these
- 00:29:52temperatures it's it's it really is
- 00:29:54corrosion
- 00:29:55driven and the the layer image is
- 00:29:57showing something similar
- 00:29:58it's an optical image rather than a um
- 00:30:01electron
- 00:30:02microscope image and and that's showing
- 00:30:04a similar thing on the c ring at the
- 00:30:05slightly lower stress there 500
- 00:30:07megapascals
- 00:30:08but we can see again multiple crack
- 00:30:10initiation sites um
- 00:30:12and uh environmental kind of marking as
- 00:30:15well going on so showing the
- 00:30:17significance of the environment to this
- 00:30:20propagation
- 00:30:22if we have a look at a few
- 00:30:23cross-sectional images um
- 00:30:25now then um we can start to see how this
- 00:30:27mechanism is interacting a bit more with
- 00:30:29the microstructure so
- 00:30:31image a there up on the the top left is
- 00:30:34jane what looks to be
- 00:30:35a bit like a corrosion pit and we've got
- 00:30:37a crack which is
- 00:30:38initiated and propagated out at the
- 00:30:40bottom of that pit we obviously can't
- 00:30:41tell what came first a bit on the crack
- 00:30:43um but we we also see the corrosion
- 00:30:47interacting preferentially with the
- 00:30:49gamma prime and this is kind of backed
- 00:30:51up by the the following edges so we can
- 00:30:52see
- 00:30:53really at the crack tip there in image b
- 00:30:55again it's interacting with the gamma
- 00:30:57prime
- 00:30:58propagating on orthogonal q flames
- 00:31:00through the material
- 00:31:01and even some bridging through the the
- 00:31:04gamut and
- 00:31:05so the gamma prime is is typically the
- 00:31:07strengthening
- 00:31:08phase or precipitate so
- 00:31:11preferentially oxidizing that is
- 00:31:13obviously locally weakening
- 00:31:15the material in a table like this crack
- 00:31:17propagation
- 00:31:21if we have a look then what we've got
- 00:31:23here are pd
- 00:31:24um plots so these represent the crack
- 00:31:27growth or propagation
- 00:31:28rate throughout the test so the way we
- 00:31:31do this
- 00:31:32and again it's quite a common technique
- 00:31:35that's used in fatigue
- 00:31:36um less so in stress corrosion cracking
- 00:31:39and again less so
- 00:31:40high temperature with deposits to
- 00:31:42creation um
- 00:31:43but it's quite a useful method for us to
- 00:31:45actually look at a bit more detail at
- 00:31:47what's going on so what we do is we pass
- 00:31:49a large current through the specimen and
- 00:31:50we have a couple of threads
- 00:31:52either side of where we expect cracking
- 00:31:54to occur and as that crack propagates
- 00:31:56and we lose
- 00:31:57specimen cross-sectional area we see an
- 00:31:59increase in potential drop or resistance
- 00:32:01which we can measure
- 00:32:02and that corresponds to to our correct
- 00:32:04size so
- 00:32:06there's two tests here they have uh
- 00:32:09comparable test conditions and
- 00:32:11other than they they're using different
- 00:32:13deposits so
- 00:32:14so a different composition of the salts
- 00:32:17so the one on the left
- 00:32:18is uh sodium potassium sulfate based
- 00:32:20salt we can see
- 00:32:21an incubation period so the crack hasn't
- 00:32:24started straight away
- 00:32:25but then it initiates and it's kind of
- 00:32:28continuing on then
- 00:32:29at a very linear rate um not really
- 00:32:32showing much
- 00:32:33correlation to to the crack size as we
- 00:32:35would expect to see with fatigue
- 00:32:37fairly linear rates trembling along if
- 00:32:40we have a look at the
- 00:32:41chloride salt we see something similar
- 00:32:42except we've not got that incubation
- 00:32:43time we do have an overall
- 00:32:45linear growth rate there's a little bit
- 00:32:47of jumping we can
- 00:32:48see going on there but i think the key
- 00:32:50thing to note here
- 00:32:51is that with these two salts we see a
- 00:32:53very different rate of propagation
- 00:32:55so obviously these aren't calibrated
- 00:32:57we've just got potential difference here
- 00:32:59but it does
- 00:33:00it does correspond to the the crack the
- 00:33:02crack size so we can see that the rate
- 00:33:04of increase of
- 00:33:04potential drop or the rate which central
- 00:33:06drops increasing is twice
- 00:33:08as high for sea salts so that chloride
- 00:33:11based so you know these temperatures on
- 00:33:14on these alloys
- 00:33:14so it's kind of highlighting the
- 00:33:16significance of of what the deposit is
- 00:33:18actually made of
- 00:33:19and or what the composition of the
- 00:33:21deposit is
- 00:33:22you know for this mechanism
- 00:33:25so starting to introduce fatigue into
- 00:33:28the equation then so
- 00:33:29now we've got stress cycling going on
- 00:33:31this is quite a simplistic view of
- 00:33:33of how corrosion or this type of
- 00:33:36corrosion affects
- 00:33:37the life of these alloys and again it
- 00:33:39doesn't tell the whole story but it does
- 00:33:41just start to help us unpick a little
- 00:33:42bit what's going on
- 00:33:44so we've got three um curves on this on
- 00:33:47this stretch cycles to fairly aggressive
- 00:33:49an sn curve
- 00:33:50at the first two or the square in the
- 00:33:53triangle
- 00:33:53are five microgram um fluxes so that's
- 00:33:57the rate of deposition
- 00:33:58of our corrosive species and the circle
- 00:34:02is 1.2 micrograms centimeter
- 00:34:06squared per hour rate of of flux
- 00:34:09deposition
- 00:34:10so if we compare the two five microgram
- 00:34:12results to start with
- 00:34:13one is a one second dwell so this is a
- 00:34:15trapezoidal test
- 00:34:16so we're ramping up the stress and we're
- 00:34:19holding it that dwell time represents
- 00:34:22the holding the stress at maximum value
- 00:34:25and then we're ramping it back down and
- 00:34:28so we've got a one second dwell in the
- 00:34:3060 seconds so the 60 seconds we'll test
- 00:34:32the triangle uh
- 00:34:34contains or has a lot more corrosion
- 00:34:36occurring per cycle
- 00:34:38than the one second twelve so we can see
- 00:34:40more corrosion
- 00:34:41in that case worse for t life fairly
- 00:34:44simple
- 00:34:44fairly simple story and if we compare
- 00:34:47the two one-second dwell tests
- 00:34:48so the square in the circle again the
- 00:34:50square
- 00:34:51has a higher flux and more corrosion
- 00:34:54than the circle
- 00:34:55and we can see again that drop in
- 00:34:56fatigue life corresponding to that
- 00:34:59increased amount of corrosion so that's
- 00:35:02all makes all makes sense but again this
- 00:35:04is quite a simplistic view so we'll
- 00:35:06we'll start to dig into a bit more what
- 00:35:08what's going on because i don't think sn
- 00:35:10curves are a great way to
- 00:35:11to look at look at what's going on
- 00:35:15so what we've got here then is uh is
- 00:35:17again using that potential drop
- 00:35:18technique but this time on
- 00:35:20fatigue specimens uh 550
- 00:35:23degrees c and these are not fatigue
- 00:35:24specimens um
- 00:35:26we've got some plots of the crack
- 00:35:29propagation
- 00:35:30um so showing how that crack depth is
- 00:35:32varying over
- 00:35:33both cycle and uh and time and
- 00:35:37and uh this sort of starts to give us a
- 00:35:39bit more information on what's going on
- 00:35:42so what we can see is uh if we if we
- 00:35:45start by looking at the cracked
- 00:35:47cycles so that put on the left on the
- 00:35:49left uh we've got
- 00:35:50three three tests notched one knots two
- 00:35:52and not just three so
- 00:35:54knots two and notch three uh are a
- 00:35:56longer dwell again
- 00:35:57than notched one so knots two and notch
- 00:35:59three have more corrosion going on per
- 00:36:00cycle and again we can see that
- 00:36:03reflected in
- 00:36:04in the fact that they take fewer number
- 00:36:06of cycles to fail
- 00:36:07uh than notch one so more corrosion
- 00:36:09fewer number of cycles
- 00:36:10fail again that same story coming
- 00:36:12through we can also see though
- 00:36:14that we get these crack jumps occurring
- 00:36:17uh
- 00:36:18as we resolve the specimen so the
- 00:36:19crosses there on on those curves
- 00:36:22represent the point at which we've
- 00:36:23resulted the
- 00:36:25the test so as our corrosive species or
- 00:36:28deposit
- 00:36:29is is used up it evaporates or you know
- 00:36:31it's it's becoming used in the corrosion
- 00:36:33process
- 00:36:34and we have to kind of replenish that so
- 00:36:37we can see how the crack jumps and
- 00:36:39interacts with with those resulting
- 00:36:41processes if we have a look at how how
- 00:36:44this works out on a time basis rather
- 00:36:46than a cycle basis
- 00:36:48um you can see the the story is a little
- 00:36:50less clear
- 00:36:51so notched 3 and notch two again have
- 00:36:54some fairly big jumps
- 00:36:55early on we've got more stress corrosion
- 00:36:57cracking going on
- 00:36:58um but they've not increased it
- 00:37:01there's a higher rate because there's
- 00:37:03less fatigue elements going on this is a
- 00:37:05much one specimen so
- 00:37:07there's no clear correlation in terms of
- 00:37:09time to failure
- 00:37:10in in this case because i think the
- 00:37:12mechanism is a bit more complicated than
- 00:37:14the sf curve sort of depicts
- 00:37:20if we move on to have a quick look at
- 00:37:22the fracture faces
- 00:37:23of those kind of specimens um
- 00:37:26we see lots of environmental markings
- 00:37:29and things going on
- 00:37:30as well which also represents um
- 00:37:34those kind of start stop type behavior
- 00:37:36of the crank propagation
- 00:37:38mechanism i think again is it's fairly
- 00:37:40complicated we've got multiple cracks
- 00:37:42interacting with each other things
- 00:37:43um but it's it's sort of useful to note
- 00:37:46that
- 00:37:46that start stop behavior is it's sort of
- 00:37:50represented on the fracture phase as
- 00:37:52well so
- 00:37:54i think quite interestingly um if we
- 00:37:56start to to look at
- 00:37:57corrosion and fatigue interactions on on
- 00:38:00a paris curve
- 00:38:01we can we can understand a bit more
- 00:38:03what's going on so i've included here
- 00:38:05on the left a paris curve and if if
- 00:38:08you're not familiar with it
- 00:38:10i'll just quickly run over what what
- 00:38:12paris curve is
- 00:38:13um so we've got stress intensity factor
- 00:38:16on the x-axis
- 00:38:18so that's measured in megapascal's root
- 00:38:20meter
- 00:38:21so it's a kind of stress to geometrical
- 00:38:25um parameter and it's it corresponds to
- 00:38:29crackdown
- 00:38:30so as our crack depth increases our
- 00:38:32stress intensity factor
- 00:38:34uh typically increases as well um on
- 00:38:37the y-axis we've got the crack growth
- 00:38:40rate so d
- 00:38:40n d a sorry over d n uh where d a is
- 00:38:44changing crack length
- 00:38:45over the end is changing in cycles
- 00:38:48and what we see for typical fatigue
- 00:38:52if we look at the red line is we see
- 00:38:54initiation so
- 00:38:56rapid increase um
- 00:39:00corresponding between stress potential
- 00:39:02or or rapid increase between
- 00:39:03uh relationships between stress
- 00:39:05intensity factor and crack growth rate
- 00:39:07and then that levels off so we get is
- 00:39:09this flat
- 00:39:10um line on the log log plot which
- 00:39:13represents a paris curve and
- 00:39:15we've got a constant and an exponent to
- 00:39:18to help us
- 00:39:19to model that uh so that's our kind of
- 00:39:21typical fatigue area and we'll use those
- 00:39:24constants
- 00:39:24uh quite frequently to model fatigue
- 00:39:28and then again just towards or just
- 00:39:30towards the end we'll see
- 00:39:31uh crack growth deviate from from that
- 00:39:34paris line
- 00:39:35um again so really that bit in the
- 00:39:37middle that cm
- 00:39:39straight line is our paris growth and
- 00:39:42that's when we know
- 00:39:43that uh our crack rate is is fairly in
- 00:39:45line with a typical fatigue dominated
- 00:39:47crack group so if we have a look there
- 00:39:49our three results plotted on
- 00:39:52paris curve so the plot on on the right
- 00:39:54here
- 00:39:55what we see if we start looking at
- 00:39:57notched one so
- 00:39:58to the left of what i put in there is a
- 00:40:01threshold line
- 00:40:02we see there's very little uh or no
- 00:40:05relationship really
- 00:40:06between stress intensity factor and the
- 00:40:08adm um
- 00:40:10early on in the correct growth of
- 00:40:13our crack propagation is is driven
- 00:40:15really more by
- 00:40:16time dependent or corrosion uh driven
- 00:40:18factors rather
- 00:40:20rather than stress intensity factor
- 00:40:22however there becomes a point at the
- 00:40:24crack depth
- 00:40:24at which the crack then starts to grow
- 00:40:26more in line with fatigue
- 00:40:28so when we look to the right of that
- 00:40:30threshold line for the notch one
- 00:40:31specimen we see
- 00:40:32we actually start to get that straight
- 00:40:34line so that paris curve
- 00:40:36um is starting to come through there so
- 00:40:38we can we can sort of hypothesize that
- 00:40:40um at that point fatigue has become the
- 00:40:43dominant factor so
- 00:40:44to the left we've got stress corrosion
- 00:40:46cracking or time dependent
- 00:40:48uh um mechanisms dominating our crack
- 00:40:51growth and
- 00:40:52to the right we've got more fatigue
- 00:40:53coming into play and
- 00:40:55i guess what's quite interesting is to
- 00:40:56compare um
- 00:40:58the knox one to the notch three uh sorry
- 00:41:01knots two and notch three
- 00:41:02specimens so not too much three or a
- 00:41:04longer dwell so there's more corrosion
- 00:41:05going
- 00:41:06on but we actually see the crack depth
- 00:41:08at which they
- 00:41:09start to behave more like fatigue or
- 00:41:13they start to
- 00:41:14correspond to fatigue crack craters is
- 00:41:16higher so that's telling us that
- 00:41:18corrosion in this case is actually
- 00:41:20suppressing
- 00:41:21fatigue and again we can hypothesize um
- 00:41:24that that's due for
- 00:41:25for could be due to a couple of reasons
- 00:41:27uh so there's
- 00:41:28correct closure uh which can occur so
- 00:41:31that that's where your corrosion
- 00:41:33product actually wedges open your crack
- 00:41:36effectively reducing
- 00:41:37um your delta k so your stress intensity
- 00:41:40factor
- 00:41:40factor range because there's less um
- 00:41:42crack open
- 00:41:43crack opening closure occurring um or
- 00:41:47there's also
- 00:41:47crack crack uh tick blunting effects
- 00:41:49which which can also happen
- 00:41:51um but i guess this relationship of
- 00:41:53corrosion suppressing fatigue great
- 00:41:55is perhaps slightly uh counter-intuitive
- 00:41:58and
- 00:41:59i guess it's important to to sort of
- 00:42:01higher in fact it's suppressing fatigue
- 00:42:03grades which is
- 00:42:04certainly not suppressing the corrosion
- 00:42:06uh or stress corrosion cracking good
- 00:42:08driven uh propagation gain going on
- 00:42:10earlier
- 00:42:11um i suppose as well interesting to note
- 00:42:14is um
- 00:42:15this is uh this is also quite similar to
- 00:42:18and the behavior of single crystal
- 00:42:20surprise uh between
- 00:42:21air and vacuum um so we will actually
- 00:42:24see
- 00:42:25um that crack fatigue propagation in
- 00:42:28single crystallized
- 00:42:29in air um is slightly
- 00:42:32worse or there's less fatigue
- 00:42:36propagation in air than there is in
- 00:42:38vacuum at high temperatures
- 00:42:40so i i guess there's some comparisons
- 00:42:42that can be made there and
- 00:42:44i suppose we see those effects more
- 00:42:45maybe in single crystal always because
- 00:42:46we don't have
- 00:42:47grain boundary oxidation going on uh
- 00:42:50forms of propagation occurring
- 00:42:56so i was going to lastly talk a little
- 00:42:58bit then about
- 00:43:00why why does all of this happen and what
- 00:43:02is that what is the mechanism behind it
- 00:43:04so
- 00:43:05i was going to start off here i've got a
- 00:43:06few um sem images
- 00:43:09um of a statically legislation again so
- 00:43:11no fatigue
- 00:43:12in these last few slides that i'm going
- 00:43:15to present you
- 00:43:16um so what we can see again in
- 00:43:20echo is what i was talking about earlier
- 00:43:21we can see the crack propagating
- 00:43:24on q planes so 10801a
- 00:43:28and we can see corrosion interacting or
- 00:43:31prevalent preferentially attacking the
- 00:43:32gamma prime
- 00:43:33um and image c there's showing
- 00:43:37a what looks to be an early corrosion
- 00:43:39pit with again
- 00:43:41corrosion interacting preferentially
- 00:43:42with the gamma prime
- 00:43:44image d um is actually showing slightly
- 00:43:47the opposite so that's a corrosion pip
- 00:43:48but it was taken on a compression side
- 00:43:50and but we can actually see that in that
- 00:43:52case the corrosion is interacting more
- 00:43:54with the gamma matrix um potentially
- 00:43:58suggesting
- 00:43:58uh you know there's some stress factors
- 00:44:02in this preferential attack um also just
- 00:44:04highlighting
- 00:44:05uh the complexity of this this corrosion
- 00:44:08mechanism as well
- 00:44:09um and certainly that echoes a lot of
- 00:44:10what i've seen and
- 00:44:12you can you can you can convince
- 00:44:15yourself of one thing
- 00:44:16and then you know several weeks later
- 00:44:17you see something that's just completely
- 00:44:19the opposite
- 00:44:19really throws a spanner in the works
- 00:44:23and so you're looking down at some
- 00:44:24higher mag images so these are
- 00:44:26transmission electron microscope
- 00:44:28images of the crack tip
- 00:44:32what we can see is the crack cutting
- 00:44:34through and propagating through the
- 00:44:35camera prime
- 00:44:36quite interestingly because because
- 00:44:37again just to highlight those those
- 00:44:39gamma prime
- 00:44:40uh phases are strengthening so they're
- 00:44:43quite a lot stronger than the camera
- 00:44:45and we can also see dislocations so
- 00:44:47stacking faults
- 00:44:48on the octahedral plane so one-on-one
- 00:44:50stacking faults
- 00:44:51ahead of the crack chip and again that's
- 00:44:54not unexpected at these temperatures we
- 00:44:55would expect to see slip occurring
- 00:44:57on those planes and but
- 00:45:00interestingly the crack doesn't really
- 00:45:02appear to interact with it so i think if
- 00:45:04this mechanism was fatigued driven or or
- 00:45:07plasticity was particularly important we
- 00:45:09perhaps see the crack
- 00:45:10interacting and propagating a little bit
- 00:45:12more on those planes
- 00:45:13than we do and but we don't it seems to
- 00:45:16just carry straight on ahead
- 00:45:18on on the cube plates
- 00:45:23so some edx then uh just to look at um
- 00:45:26what's going on and again i've just
- 00:45:29taken a little snapshot i've looked at
- 00:45:31quite a few of these but what we can see
- 00:45:34is um
- 00:45:35firstly there's no convincing evidence
- 00:45:37of diffusion or absorption
- 00:45:39of corrosive species ahead of the crack
- 00:45:41tip it all
- 00:45:42appears to be happening at the
- 00:45:44corrective there's not much going into
- 00:45:45the material
- 00:45:46and certainly not a lot of evidence i've
- 00:45:48seen of that
- 00:45:50we do see presence of electrolytes at
- 00:45:52the crack tip so you can see we've got
- 00:45:53sodium
- 00:45:54uh there right down at the crack so top
- 00:45:56left
- 00:45:57uh if we have a look along we've got our
- 00:45:59allying elements and you can see how
- 00:46:01they segregate between the camera and
- 00:46:02the camera prime
- 00:46:03uh quite interesting nothing too
- 00:46:05controversial there other than the
- 00:46:06tungsten
- 00:46:07uh in orange uh which we've found in the
- 00:46:09gamma prime and i think that's
- 00:46:12slightly debated but certainly we see it
- 00:46:14we've seen it
- 00:46:15more importantly uh we've also got
- 00:46:18oxidation going on right down the crack
- 00:46:20so
- 00:46:20in green in the bottom left we can see
- 00:46:23uh our oxidation is happening
- 00:46:25right down right down at the crack
- 00:46:30so the proposed kind of mechanism i've
- 00:46:32caught with that is uh
- 00:46:34sort of fairly similar to stress
- 00:46:37corrosion cracking
- 00:46:39it certainly has a few uh analogies
- 00:46:42which could be made to stress corrosion
- 00:46:44cracking aqueous stress growing and
- 00:46:45cracking at lower temperatures but i
- 00:46:47suppose there's a high temperature
- 00:46:48uh mechanism and and while stress
- 00:46:52corrosion cracking i think is is a bit
- 00:46:53better understood
- 00:46:54i think we're still really trying to
- 00:46:56understand in some
- 00:46:58detail the chemistry of what's going on
- 00:46:59here so this is a
- 00:47:01simplistic view of my my hypothesis
- 00:47:04um but what we see is typical uh
- 00:47:08hot corrosion type features so we've got
- 00:47:10a nickel cable oxides
- 00:47:11uh below that we've got dissolved oxides
- 00:47:14and then below that we see kind of
- 00:47:18an electrolyte um and i've put um
- 00:47:21sodium sulfate there but uh i think
- 00:47:24again
- 00:47:24there's there's a number of compositions
- 00:47:26that i like my electrolyte could be
- 00:47:29so we then see formation of a pit or a
- 00:47:31crack
- 00:47:32uh kind of locally and again
- 00:47:35i don't think there's any um definitive
- 00:47:38way i don't think it's a pit then a
- 00:47:39crack or a crack kind of tip
- 00:47:40i think it kind of probably just depends
- 00:47:42a lot of the time
- 00:47:44and if we move down then to image b um
- 00:47:46in terms of how this
- 00:47:48sort of um electrochemical circuit gets
- 00:47:51set up in the crack
- 00:47:53so we see our electrolytes wicking down
- 00:47:55the crack and we see oxidation happening
- 00:47:57kind of around the crack the cracked tip
- 00:48:00and we'll see a localized anodic region
- 00:48:01region somewhere near or at the crack
- 00:48:04tip but
- 00:48:05really that's where we've got sort of
- 00:48:07acidic conditions and dissolution of the
- 00:48:09oxide
- 00:48:09and the substrate going on and i think
- 00:48:11it's that localized nature really that
- 00:48:13drives
- 00:48:14this this mechanism we don't see
- 00:48:16broad-fronted attacks we see
- 00:48:17very localized attacks which cause you
- 00:48:19know colonies of cracks
- 00:48:22and and then the image on the right i
- 00:48:23suppose is is me trying to
- 00:48:25theorize as to why we see corrosion
- 00:48:29cracks or corrosion driven cracks
- 00:48:31propagating on orthogonal
- 00:48:32uh planes or the q-flames rather than
- 00:48:35where we would typically see them on
- 00:48:37uh you know the the the slip flames the
- 00:48:39one one one of the hedral flames
- 00:48:40and and i think that's because if you
- 00:48:42take a point uh at the cracked tip
- 00:48:44and where you have a low-priority region
- 00:48:48and you grow that out you can see that
- 00:48:50the the gamma prime features which is
- 00:48:52kind of
- 00:48:52or the gamma prime phase which is going
- 00:48:54to preferentially interact with
- 00:48:56uh from that point are all kind of
- 00:48:59orthogonally located uh from from that
- 00:49:01crack tip so
- 00:49:02so that's kind of my theory as to one of
- 00:49:04the reasons why we see this preferential
- 00:49:07orthogonal growth propagation occurring
- 00:49:12so i think with that hopefully i've not
- 00:49:13bought you too much
- 00:49:16but yeah thank you thank you very much
- 00:49:18for listening
- 00:49:19and welcome any questions
- 00:49:31okay so thank you very much for that
- 00:49:33laurie um
- 00:49:34a really really fascinating uh walk
- 00:49:37through some of some of the some of the
- 00:49:38detail there
- 00:49:39on on on what we're doing
- 00:49:43in terms of physics modelling for for
- 00:49:44corrosion fatigue
- 00:49:46um there's a couple of questions come in
- 00:49:49actually there's one particular question
- 00:49:50that's coming
- 00:49:51um about about um connection problems
- 00:49:54and i don't know if
- 00:49:55if that's been a very localized issue or
- 00:49:57if that was something
- 00:49:58that that a number of people have seen
- 00:50:00there uh all i can do really is
- 00:50:02apologize for that at the moment
- 00:50:04um obviously the anarchy will get in
- 00:50:06touch with yamaki
- 00:50:07and talk to them and find out find out
- 00:50:09whether or not there is an issue
- 00:50:11with with the platform although it's
- 00:50:13just a localized issue with that
- 00:50:14particular individual but
- 00:50:15like so if if there has been an issue
- 00:50:17for you then then uh we're very sorry
- 00:50:19about
- 00:50:19it um and uh and i guess it's obviously
- 00:50:22i'm a key to to investigate what's going
- 00:50:23on with this with this
- 00:50:24platform i think it's work cast is the
- 00:50:26platform that we're using today
- 00:50:28um so moving on from that we have got a
- 00:50:31bit of time for some questions and a
- 00:50:33number of questions have been coming in
- 00:50:35um i'm gonna i'm just gonna
- 00:50:38i've been i've been sifting through them
- 00:50:40a little bit to trying to try and um
- 00:50:42to try and get some essential ones out
- 00:50:44but actually there's a load that'll just
- 00:50:45come in in the last couple of minutes so
- 00:50:46i'm going to kick off with a question uh
- 00:50:49directed to you laurie and then i'll
- 00:50:51then i'll have a look through the inbox
- 00:50:52and see if there's any other interesting
- 00:50:54stuff that's coming
- 00:50:55so the first question from um arun
- 00:50:59um can't work out where your from a roon
- 00:51:02you're an engineering manager the email
- 00:51:05address is msn.com
- 00:51:07the question is how do you assess
- 00:51:09standstill corrosion in a gas turbine
- 00:51:12it's a very good question so lori i
- 00:51:14don't know if you've got any thoughts on
- 00:51:15that
- 00:51:17yeah thanks john um i'm just standstill
- 00:51:21corrosion isn't isn't
- 00:51:22a term i'm necessarily familiar with but
- 00:51:24i can sort of
- 00:51:25imagine you're referring to corrosion
- 00:51:29kind of you know not during operation i
- 00:51:30don't know what your view is on that
- 00:51:32john yeah that's that's exactly what it
- 00:51:36is so
- 00:51:37obviously yeah if you leave a if you
- 00:51:38leave a machine uh standing still
- 00:51:41uh especially uh near a marine
- 00:51:44environment
- 00:51:44um how do you protect that from from um
- 00:51:48from corrosion and the answer is um you
- 00:51:50protect it from corrosion by
- 00:51:52by blocking the airways and keeping it
- 00:51:54warm and keeping it ventilated and
- 00:51:56keeping it dry
- 00:51:57um that's not something that would
- 00:51:58particularly cover
- 00:52:00directly in what we're doing but you
- 00:52:02know that's not to say that we can't
- 00:52:04develop corrosion models
- 00:52:06that are appropriate especially for the
- 00:52:08compressor and all sorts of components
- 00:52:10um the answer is we we can develop those
- 00:52:13models and we can put that physics into
- 00:52:15into into into these models but
- 00:52:18primarily what we're doing is looking at
- 00:52:20looking at what's happening during the
- 00:52:21operation and and whether or not you can
- 00:52:24get more life or get more use from your
- 00:52:26engine rather than
- 00:52:28get more standstill time out of your
- 00:52:29engine um i think it is a short answer
- 00:52:32but the answer the long answer is we we
- 00:52:34can develop we can develop those methods
- 00:52:36if you need us to
- 00:52:39okay i don't know if you've got anything
- 00:52:41to add to that laurie um
- 00:52:42you might yeah we can't do that no i
- 00:52:46don't think you're right that was that
- 00:52:47was my kind of understanding as well i
- 00:52:48guess
- 00:52:48it's standstill corrections props
- 00:52:50perhaps a bit more of an issue now
- 00:52:51there's
- 00:52:52lots of assets not being used due to
- 00:52:54code
- 00:52:55so uh so yeah probably quite quite
- 00:52:57relevant question
- 00:52:59okay all right so i'll tell you i mean
- 00:53:00there's a number of questions that are
- 00:53:01coming through on
- 00:53:02the general and there's actually one two
- 00:53:05that's been directed to me so i'm going
- 00:53:06to start answering some of those
- 00:53:07questions if you want to look through
- 00:53:08the inbox and see if there's anything
- 00:53:10that you want to answer there laurie
- 00:53:11then then
- 00:53:13and look at that and um i'll hand over
- 00:53:14to you in a minute yeah
- 00:53:16cheers yeah okay so there's there's one
- 00:53:19here from
- 00:53:20james maddock at rolls royce uh directed
- 00:53:22to me
- 00:53:24um you mentioned the use of predictive
- 00:53:26physics-based approach
- 00:53:28to potentially take credit for park
- 00:53:30power operation to extend the life of
- 00:53:32assets
- 00:53:33the question is do you use this for
- 00:53:35safety critical applications
- 00:53:37is your tool set qualified for this i
- 00:53:40guess the question from rolls royce a
- 00:53:41safety critical operation
- 00:53:43would be potentially an aero engine and
- 00:53:46the the answer is we haven't developed
- 00:53:48it for rolls-royce aero engines
- 00:53:50um but we do believe the physics is
- 00:53:52sound and we do believe that
- 00:53:54that we can certainly support you in
- 00:53:56your operations and i think actually
- 00:53:58some of our engineers are supporting
- 00:53:59your your your teams in terms of um
- 00:54:03physics based assessments of your
- 00:54:05components um
- 00:54:06but primarily the work that we've done
- 00:54:07is on the industrial gas turbine side
- 00:54:10and that's that's that's anything from a
- 00:54:13from a 10-ish megawatt machine uh
- 00:54:16to a to a to a full-size frame engines
- 00:54:19which are operating in power stations
- 00:54:21around
- 00:54:22uk and actually europe um and and
- 00:54:25we i mean there's something there about
- 00:54:27the the tool set you know
- 00:54:28we have we have methods that we develop
- 00:54:30but we don't have a
- 00:54:31set of software that we just license
- 00:54:33that you can plug your engine into
- 00:54:34we we we developed bespoke solutions and
- 00:54:38those solutions can be for the oem who
- 00:54:40runs a fleet of 15 000 engines
- 00:54:42or it can be a bespoke solution for a
- 00:54:45particular power station that has a
- 00:54:46particular type of gas turbine
- 00:54:48and um and we'll develop those methods
- 00:54:51either way and and they have been
- 00:54:53validated we work with the
- 00:54:55power station we work with the insurer
- 00:54:57and we work with the oem
- 00:54:59to to to see the best way of of managing
- 00:55:03that plant and and quite often we're
- 00:55:04very successful in getting more hours of
- 00:55:06operation
- 00:55:07out of that plant before the before the
- 00:55:10before the maintenance window is due
- 00:55:13um hopefully hopefully that answers your
- 00:55:16question james
- 00:55:18um there's a couple of questions coming
- 00:55:22on on the uncertainty modelling and
- 00:55:25and the bayesian approaches if you like
- 00:55:28um
- 00:55:29one of them oh gosh something's happened
- 00:55:30to the order of the questions while i've
- 00:55:32been talking there
- 00:55:33one of them comes from ian mcafee
- 00:55:37you've asked how big a problem is
- 00:55:40uncertainty
- 00:55:40in measurement drift characteristics of
- 00:55:43the thermocouples what is the effect of
- 00:55:44this uncertainty
- 00:55:46and then cam i think you're in hong kong
- 00:55:50cam you've asked may i know more about
- 00:55:53risk-based decision tool
- 00:55:55um i think there's some more come in
- 00:55:57whilst i've been talking
- 00:55:58and whilst laurie was talking there on
- 00:56:00uncertainty it's quite
- 00:56:02it's quite an interesting area i think i
- 00:56:03think you know what the the the short
- 00:56:05answer is i think we need to do
- 00:56:06we need to do a session um i'm gonna i'm
- 00:56:08gonna try and organize a session
- 00:56:10with the imac e and do another one of
- 00:56:12these webinars where we can talk about
- 00:56:13the risk-based methods that we've been
- 00:56:14using
- 00:56:16um getting into the detail the the
- 00:56:19there is a massive problem with
- 00:56:20uncertainty on the measurement side
- 00:56:22um you know what what happens if a
- 00:56:25thermocouple fails whilst whilst you're
- 00:56:27sat there monitoring it how do the
- 00:56:29algorithms deal with that
- 00:56:31uh yes there's drift on thermal couples
- 00:56:32and yes there is
- 00:56:34there is inaccuracy in the measurement
- 00:56:36you know you know if you just think of a
- 00:56:38basic
- 00:56:39sort of rule of thumb type thing then
- 00:56:40you know a 10 degree difference
- 00:56:43in in in temperature can give you
- 00:56:46the order magazine of about 50
- 00:56:48difference on a creep life
- 00:56:49calculation so so you know that that
- 00:56:52level of uncertainty on the measurement
- 00:56:54can lead to quite quite a lot of
- 00:56:56uncertainty in terms of
- 00:56:58your prediction of creep damage and
- 00:57:01therefore the the the amount of
- 00:57:03damage that you've actually done to your
- 00:57:04engine um and
- 00:57:06a simple way of sort of adding all of
- 00:57:09those uncertainties up
- 00:57:10will generally if you just add those
- 00:57:12uncertainties on top of each other
- 00:57:14you will end up in a position where
- 00:57:16you've got so much uncertainty
- 00:57:17that you've got absolutely no no no no
- 00:57:20no
- 00:57:20clear view on how much damage you are
- 00:57:23predicting with your engine
- 00:57:24and so so this is this is where the the
- 00:57:26probabilistic methods
- 00:57:28uh using using um that we've been using
- 00:57:31on on nuclear power stations and
- 00:57:33actually adding up those uncertainties
- 00:57:35in the correct way to give a realistic
- 00:57:39view of your uncertainty that's that's
- 00:57:40where the value really comes into this
- 00:57:42and i think we do need to put a
- 00:57:43presentation together on how we're doing
- 00:57:44that because
- 00:57:46because it's a little bit difficult for
- 00:57:47me to answer in in
- 00:57:49in in this session but um
- 00:57:52i'm going to make that happen i'm going
- 00:57:53to make that happen we are going to do a
- 00:57:55session on that
- 00:57:56um and hopefully you guys can watch that
- 00:57:58and learn more about how we do that
- 00:58:01in the meantime if you want to get in
- 00:58:02touch i can put you in touch with those
- 00:58:04teams and they can talk about that more
- 00:58:06um hopefully
- 00:58:08that's enough on that um pradeep
- 00:58:12where are you from you're from the
- 00:58:14ministry of
- 00:58:15defense standards engineering um
- 00:58:19that's uh that's interesting so you
- 00:58:22asked an interesting question
- 00:58:24do you use machine learning and
- 00:58:25artificial intelligence in your data
- 00:58:27analysis models
- 00:58:29um not in this instance we do have that
- 00:58:32capability within the company
- 00:58:34and and we we can use that capability in
- 00:58:37terms of trending
- 00:58:38what's happening to fleet of engine uh
- 00:58:41and
- 00:58:42and that type of stuff um what we're
- 00:58:45talking about here today is really about
- 00:58:46our physics-based approach
- 00:58:48to to understanding what's going on with
- 00:58:50the damage accumulation within the
- 00:58:51engine
- 00:58:52um and there's a there's a time and a
- 00:58:54place for for those sorts of things so
- 00:58:56if you think about highly complex
- 00:58:59systems
- 00:59:00like the hot gas path within a gas
- 00:59:02turbine it's very very physics driven
- 00:59:04uh machine learning algorithms are not
- 00:59:07particularly suited for that type of
- 00:59:08stuff
- 00:59:09if you think about a situation like
- 00:59:12rotor dynamics
- 00:59:13you can predominantly you're looking at
- 00:59:16a an uncertainty model
- 00:59:18you know there's so much uncertainty in
- 00:59:20what's going on with that vibration
- 00:59:21monitoring that the physics model gets a
- 00:59:22little bit swamped
- 00:59:24from my experience i'm sure there's
- 00:59:25people that would disagree with me
- 00:59:27on that but um and i think that's kind
- 00:59:29of there are scenarios like that where
- 00:59:31machine learning and artificial
- 00:59:33intelligence can can actually be a lot
- 00:59:35more helpful than the physics-based
- 00:59:36approaches
- 00:59:37um that would be my answer i don't know
- 00:59:39if you've got any further views on that
- 00:59:40lori
- 00:59:42yeah i i'd agree with what you've said i
- 00:59:44think as well though um
- 00:59:46with with the with the physics models
- 00:59:48we'll we'll understand them and
- 00:59:49and you know generate them from testing
- 00:59:52like i've shown or
- 00:59:53um analysis um and
- 00:59:57get the model but often will then fit it
- 00:59:59to large data so
- 01:00:00it's not a typical machine learning as
- 01:00:02such but it is
- 01:00:03kind of you know an automated fitting
- 01:00:05process often i think
- 01:00:06um where you have a lot of data and you
- 01:00:08have a model but you need to look at
- 01:00:11how that you know what what your
- 01:00:13constants and parameters are
- 01:00:14for your model you know considering your
- 01:00:16data um so i think that's
- 01:00:18kind of machine learning in my in my
- 01:00:20eyes my mind
- 01:00:21it's more fitting though i suppose yeah
- 01:00:25okay i'm sure that that our answers the
- 01:00:28question and their answers is probably
- 01:00:29open to the whole
- 01:00:31area of debate there but actually
- 01:00:33looking at the clock
- 01:00:34we're pretty much out of time i think i
- 01:00:36mean i don't know if you've been looking
- 01:00:37through the inbox
- 01:00:38and whether or not you've spotted any
- 01:00:39more questions that'll come in and if
- 01:00:41you want to answer any of them before
- 01:00:42before we close out
- 01:00:46uh yeah i think you're right we're
- 01:00:48running a bit low on timeout there were
- 01:00:50a couple i'm just trying to see
- 01:01:00there was one there was one by alfred on
- 01:01:03single crystal blades being good for
- 01:01:04high temperature but not for
- 01:01:06h2s and co2 environments and i think i
- 01:01:09made made mention to this that yeah as
- 01:01:11we
- 01:01:12try to sort of push the operational
- 01:01:14temperatures of these we turn more and
- 01:01:16more to single crystal blades for their
- 01:01:17their creep properties i think in
- 01:01:19particular um but often yeah
- 01:01:21the detriment of the corrosion and um
- 01:01:24you know oxidation
- 01:01:25properties um so i think
- 01:01:29there's we can obviously look to
- 01:01:32different materials but i guess coatings
- 01:01:33as well is another
- 01:01:34another area and several surface
- 01:01:36treatments and
- 01:01:38offer possible solutions that you can
- 01:01:40you can look to um
- 01:01:41but yeah again it's obviously all all
- 01:01:43quite uh sort of
- 01:01:45complex stuff and needs individual
- 01:01:47consideration i think perhaps a lot of
- 01:01:49time
- 01:01:50yeah i agree with that actually i mean
- 01:01:51so that's a really good point by alfred
- 01:01:53um you know that certainly the the newer
- 01:01:56generation of single crystals which have
- 01:01:58got
- 01:01:58quite low chrome contents uh they tend
- 01:02:01to
- 01:02:01suffer more in these types of
- 01:02:03environments than the earlier generation
- 01:02:06and um and actually we we we've um
- 01:02:09we we funded a phd at cambridge
- 01:02:10university just look at that problem in
- 01:02:12particular
- 01:02:13and we never got to the techno economic
- 01:02:16assessment for one of the better word
- 01:02:18which which sort of tries to weigh up
- 01:02:20the argument of
- 01:02:21is is it better to put an expensive
- 01:02:23coating on or is it better just to use a
- 01:02:25a sort of an earlier generation of
- 01:02:27single crystal which has got higher
- 01:02:29chrome content
- 01:02:30it's an interesting subject area and
- 01:02:32i've never got that
- 01:02:34deep in it but it's a really good point
- 01:02:35by alfred i think
- 01:02:37yeah definitely yeah okay
- 01:02:40so i think i think i mean we're running
- 01:02:42over the clock and people have probably
- 01:02:43got
- 01:02:43meetings to get to on teams and that
- 01:02:45type of stuff so
- 01:02:47i think i think we're going to pull
- 01:02:48stumps there unless unless there's any
- 01:02:50other burning questions that you want to
- 01:02:51answer laurie
- 01:02:53uh now i'm happy with that thanks john
- 01:02:55all right well then i guess
- 01:02:56i guess it's time just to say thank you
- 01:02:58very much to everyone who's made it this
- 01:03:00far
- 01:03:00and i hope you found it interesting and
- 01:03:02um like i said
- 01:03:04i'll we'll try and get some more
- 01:03:06sessions organized around the
- 01:03:07probabilistics
- 01:03:08and and that type stuff and hopefully
- 01:03:10speak to you again soon
- 01:03:14okay goodbye
- gas turbines
- sustainability
- asset management
- physics-based models
- Fraser Nash
- engineering consultancy
- high temperature materials
- single crystal super alloys
- digital assets
- creep and fatigue