Sin Sin Hsu – S&C Performance & Reliability

Just another WordPress site

Sin Sin Hsu – S&C Performance & Reliability

hello today I’m presenting some modeling work that I’ve done to improve their S&C performance and reliability within that work rail mmm what I found after joining the SNC team two years ago actually is two years ago today now join us and see team and what I found is that we may design a component or layout that we think is perfect but if it excites the resonance frequency of the train or causes unnecessarily large forces then it’s not a good design so I’m modeling and data analysis tools that I use is some vampire and track ax so these are vehicle dynamics simulations packages track axes in house of metal rail in-house software based on vampire it runs on Excel so and anybody can use it so a vehicle dynamic simulation this is some developed this was developed by British Rail in the 70s and 80s it seems that the train is made out of masses and springs and dampers so it’s a rigid rigid body motion method and then it has a contact patch share modeling and then the track model as well so you combine everything you can do you can calculate the response of a vehicle and also the forces at the contact patch so the forces will detect dictate the degradation of your track so that’s very important so this example is an animation from vampire I got it from Delta Rho at eliminate minus– thank you so it shows them this shows them the train running through the crossing so the forces can see a quick jerk as it goes through the crossing so it shows the forces at the wheel there so you can see arrows coming down up and down under the wheel that’s the forces acting on the rail so you can use it to calculate forces and the response of the vehicle and see if the designers so other tools are the cat and finite element modeling tools SolidWorks I use it to account fake distress on components and also to do some assembly motion analysis so the picture there show some an example of my work I was looking at the effect of lipping on the stop rail that’s acting on the switch rail so if you push the switch well what is the switch real seen so this is one example and this is the switch flexure model that I’m looking at so I can put a force anywhere along the switch plate and see what the reaction forces and the back of the switch and also what the stress level is at the cutout at the foot I’m also looking at the supplementary back ride this is a model our firm a supplementary back track or CV s switch so the graph there shows the reaction force on one of the crank faces okay so and the third software package that I use a lot is mat lat it’s a let’s say like a high-level programming data analysis package so I use it for collecting data plotting graphs and processing large volume of the measured data so the graph that shows some real profiles that I measured on track I can have five hundred real profiles and I just wait write a program in MATLAB and you read the five hundred profiles within a few min then plop them out and then the dots there are the contact position by a wheel so blue and red shows blue is the design chrome biles red are the measured profile so you can compare the difference between the design and the measured profile so I’ll show a series of examples that I’ve done using using all these modeling techniques so example one is and attract design optimization I do a lot of these work with northern hub with David David woods and Crossrail and the roots as well so when they have a new layout they want to know if the layout has been optimized because in track design you have a wide range of parameters you can use but is it the best optimized values that you’re

choosing so so my modeling using vampire is to even out the where and these are CF damage which is rolling contact the teeth at the all the wheel sets to avoid high localized wear because if you have high localized Square and RCF you need to change the realm so the whole life Kostis becomes quite high okay so this is an example that I’ve done with David it’s also Lane proposed option 20 so there are six routes there the oven down chat Moss up and down Bolton and two siding tracks so from the simulation the output would be the where so that’s the energy transfer to derail set that shows the wear rate of on the rail and then you have the RCAF damage so would you have cracks and all that if you have this design so I use the traffic light color coal to present the predicted severity of the damage so it’s they’re green and and that color coat so for example down chat mas I was given this you have all the elements the speed the radius and the applied Kent and the can’t deficiency so I did calculations you my knight in shining armor so this is the example that I did I said the where calculation is either there are CF calculations I looked at both the low well and the high well and they’re clenched the gauge corner as well and I proposed changes to the design if I can’t change anything I’ll propose maintenance regimes so it’s for example this part of the rail will experience a lot of RCF problems then whose premium rail for this part and if it’s varying too much put some raluca craters here so I proposed things for different sections of the tracks okay so recommendations to lubrication can’t deficiency usually and maximize the can’t deficiency because increasing tenth efficiency is good because the Train would be more radio when it’s curving so more radio means you you exert less less force on the high realm okay so more even force not at the leading wheel and then you preserve the lower low low rail because the low rail will suffer from plastic flow corrugation if you’re over counted so with those problems you can’t do anything with plastic flow you can mitigate it so if you increase can’t efficiency you say low well so and the result from all these study is that it’s easier construction because we have no account it helps with the gauging issue as well with low account and also we reduce aware Nasia damage on the rail and so therefore less maintenance and lower whole life cost for the whole layout so this is the second example I’m presenting its new switches at risk of causing derailment I was asked to go out on site two years ago to help hold back junction in Leeds because a new set of evie switch has failed the inspection the inspection means that you put a tgp eight gauge on which is some one shown on the right that’s a line there that line shows sixty degrees so if the contact position falls anywhere below the line it means that the contact angle is less than 60 it is less than 60 that’s a risk that the train will derail so this is our inspection regime and this new switch fail inspection so what is the cause of it so I did some real rail interface analysis and found that and took measurements of the profiles and I and then I used vampire to look at it and indeed the contact angle is 56 degrees which is the action would be to ban or facing traffic so that’s a lot of delay minutes and cost so investigating this case we found that the root cause was because of the loose machining tolerance pacified in the relevant standard the standard was 400 for so what we then do is to change the standard we issue a letter of instruction 283 to all S&C manufacturers

in February 2013 and we tighten three main dimensions because there’s so many machining pole machining tolerance on the switch if you tighten everything you’ll be so expensive you won’t be able to make the switch so we only target three of them the key mentions so if the targeted three key dimensions the contact angle will not be less than 58 degrees so one example of the dimension was the gauge corner radius the radius is 12 mil but the tolerance was plus minus 2 mil so that gave 5 degrees that is the main reason but we change it to point two five mil plus minus 0.25 mil we could do that because nowadays they CNC machined rails not planing it okay so that that might as to Mel was like 20 years ago so we also found that this loose machining tolerances and in fact on switch plate damage so this figure here shows that using those measured profile I run a vampire analysis and show that with design profiles the last contact on the switch plate in the trailing Direction is 600 mil from the toe at that position the switch rail is 8 mil thick so it’s good but with the measured profile I run the same simulation and the contact position is actually at 140 milk from the pole and the width of the stretch is 4 mil so that’s why sometimes we have switch plate breaking off because of the tolerance issues so loose machining tolerance can also cause premature switch plate damage so the result is this finding we have now reduced a derailment risk with new switches and then the fewer so5 3 inspection failures because of that and associated delay minutes by banning the trains and also the remedial grinding work that has to be carried out and also we reduce incidents of amateur switch plate damage so example 3 is the switch topping design and derailments we have at least six different switch topping designs in our suite off seize the topping design means the relative height between the stock rail and the in and switch realm so what what we found is that when we ask people to go and repair the switch the grinders go out and it’s not grinding the switch down there’s no limit where how much they can grind to so they keep grinding it until the topping depth is quite low so when is their what is done say that until a wheel climb wheel can start climbing up the switch what is the relevance of this in princess Street Gardens to derailment the happen in Twitter 2011 the switch rail was very very low it was about 27 mill loan lower than their stock rail head okay so I’ve plotted all the topping designs as well as the topping at princess Street Gardens and stro spree it shows that at the the limit of 20 Mill it becomes unsafe so I’d put princes street gardens to train derailed around a topping depth of 23 throws free it derailed at around topping depth of about 21 mill so there is a limit so we’re setting a new grinding limit if you find the switch topping being 20 Mill or below 20 Mill or deeper then that’s it no more grinding you have to well repair to build up the switch so this is published in a new letter of instruction last week so in the topping design again we have this straight switch straight type switches as you can see is just a straight line and a straight line going out these are the old switches so we had a couple of derailment on these switches and we found that actually we have three different topping designs by three manufacturers so we need to standardize a topping design to make it safe so I did some vampire analysis and assess the where and derailment risk so these are the joys REE PW drawings for these switches it’s in done in by hand so you have like notes routing from the drawings don’t uses use that so there are some different interpretations of these drawings by the three manufacturers so I did some analysis and we standardized it as a progress rail

topping so we will reduce the risk of derailment and wear so the result from this example is that a better understanding of switch topping designs and therefore able to reduce the risk of derailment by changing standards and joins example for switch grinding practice as we say as we share as we saw previously new switches people are going out to grind it down and all that and I realized that the grinding process is that you have to grind down the switch to the design blade angle which is 76 degrees so this switch here has contact angle of about 44 degrees so together to 76 degrees which is specified by a standard you need to grind off at least five mill of the switch so that is effectively reducing a lot of them reducing a lot of material after switch just to make it compliant and that involves 30 grinding passes so it takes a whole night just to grind it down so I did some analysis and then I said okay let’s change the practice we’re grinded to 65 degrees instead minimum of 65 degrees so I did another analysis it’s all fine 65 degrees 60 degrees is still limit 65 degrees has some factor of safety in it so and also we only target twenty mill deaths down to 40 mil death because the own 40ml the wheel will not see it so that will reduce a lot of work being done like rush job by the grinders trying to get everything done and then open to traffic so that’s no more rushing around just make sure you blend everything make sure there’s no discontinuity between the work and what what was there before so that is the requirement so this is published in a letter of instruction so using using the measured angles we came out with maintenance limits so this is again in the same letter of instruction so anything below 55 degrees contact angle just ban any traffic because they’re derailment extras free the contact angle was about fifty fifty two so even if you have very you lubricate your switches very well the wheel will still climb so anything between 55 to 60 then you can do as what the inspection standard says apply some lubrication grind within 36 hours and between 60 and 65 degrees as in new instructions try and grind within 13 weeks or monitor the wear every 13 weeks to make sure that it doesn’t go below 60 so this is our new maintenance limits so the results from this work is better guidance for maintenance therefore reduce the risk of derailment due to rush repair work and also less grinding is required so you reduce the cost and labor as well as the material you don’t have to replace that switch that often example five relaxing the FPL pollens the front point lock tolerance I was asked by the roots if we can relax the tolerance a bit at the moment it’s like at three point five mill and detection of five mill if we can increase the tolerance then we have more we can run more trains we all have that many point failures so I did some real real interface analysis because this is an important issue because relaxing the FPL tolerance could increase the risk of train derailment because the train wheel might strike this rich well it might not just climb immediately but rest of the train might climb because the switch has been flattened and creates a table running table for it to run up so I did some real real interface um analysis and I found that the switch gap is three point five milk for vertical switch switches that’s how far you can open the switch until the the wheel strikes it but that’s using our new pH switch I mean new PA wheel profile if you have a one wheel that gap will reduce because the plunge face of the wheel is super so in fact we are near to bother border border line case so I did all the analysis although with an r60 you can relax it a bit but we have one wheel you will reduce it so so in fact I don’t recommend it to be relaxed in fact should we tighten it so example says common crossing profiles

optimization our design of the common crossings are generally over 40 years and vehicle characteristics have changed in the 40s so we’re proposing some changes to the current design this is an example of a crossing at Shelford Junction we paint sprayer and see what the Train was doing across it the train was running from the left hand side to the right transferring from the wing to the nose that’s a strange gap at a nose it’s hitting the side of the nose and there’s a strange gap there so we investigated what was happening there so I’ll show you the results after but we have a problem in that with rail because so our crossings are failing between 5 and 10 years of service crossings are designed to last at least 20 years they’re meant to last for for 450 million gross tons but they’re lasting less than that at a moment so what is the reason possible possible reasons are we are increasing our traffic volume and we’re increasing in 9 speeds and we have less access for maintenance so something has to change we renew our layout configurations changing it from timber to concrete that increases the track stiffness so we don’t understand the Pats and all that enough to make it replicate what the timber was doing our crossing designs that I said shopping designs have not changed since the 70s that includes their 1 1 3 a and then our 60 topping was optimized for P 1 wheel profiles but we now use p8 and p5 so we look at our issues how big is our problem so I looked into the our DMS which is our rail damaged managed measurement and found that in one year November 2 2012 to November thirteen we had 9500 crossing defects common crossing defects and we’re struggling to close these defects in fact we closed 2,000 of them so these are the main issues top 80% of the crossing issues they’re squats on the crossing flipping problems at the nose trunks and nose and we’re of Milton wing transfer area and 14th on the list if you look at the right hand nose column 14 on the list is vertical cracks on the foot of the crossing so although it’s 14 on the list but it has great performance impact because we need to put an ER ESR on emergency speed restriction on and then we have to sometimes wait for three to four months before we can replace the crossings so that is a big performance impact so we now know that if we focus on a wheel transfer area we can actually reduce 28% of our defects or more because I don’t know where the squirrel casting they could well be at a wheel transfer area so focusing on the wheel transfer area we can we are looking at their crossing foot we’re looking at a wing rail and we’re looking at the crossing nose so the figure photos here shows on the left the wing rail indentation drill do two trailing moves on the crossing and the top right shows squat developing at the wheel transfer area on the crossing notes and the bottom figure shows the crack at the foot so crossing foot flange design we have from these scalloped design for most of our crossings cast center block ones and casts morning block crossings so we’re finding that cracks are forming where the Scout goes in so can we change that design we have continuous full flange design used forecast and the block crossings wider than one in 15 P I think this was because we used to use rail sections just bolted on hence they have continuous foot finders so perhaps we should specify continuous foot flanges for all crossings not just for selected few so it will increase the stiffness at the foot region but if you flip the crossing over you can see that there’s a

massive section with lateral material there so where do you think the crack will form so we have to look at the crossing design as well at the base whether its longitudinal rips or lateral rips so now for nose and wing topping designs we have two types of wing designs flat wing rail and inclined wing rail so the photo on the Left shows the flat wing rail design so the wing rail is effectively flat and inclined wing rail design has a slope one in 20 so will contact the crossing at crossing nose if you look at the graph on the left that’s the flat wing rail design you can see that the wheel is less supported at the wing rail there’s very little contact there and if on a curve it actually strikes nose there’s some overlap at the nose where else on the inclined profile on the right you can do that that’s more support on the wing rail and there’s very little overlap at the nose so inclined table design seems to be better it reduced the damage to the crossing nose it provides more support on a wing rail and moves the wheel transfer area to where crossing nose is bigger behind the tip and also you have a bigger cross-section at the wing perhaps this will increase the rigidity at the foot wing rail support the amount of wing rail support is determined by the width of the wheel and the wing row design I think I was speaking to people who have worked for British Rail and all that I think the size of though the width of the wheel has been reducing over the years perhaps to reduce the unsprung mass and the material so um the graph on the Left shows them the wheel transfer for a one-in-six crossing so if you have inclined wing rail design the wheel will still transfer fully to the nose after a 100 mil from the nose so this is a summary of what I’ve said the wing rails support with sharp angle the wheel transfer is near the nose near or at the nose and that’s a large tip angle and therefore impact forces but with a flatter angle the support is provided by both the wing and the nose for a longer distance therefore there’s less damage to the nose for a given speed example of wheel transfer so these are the sites that I’ve been been to and spraying paint and all that to see what’s going on so mmm the red line shows overlap between the transfer of nose and the wing you can see that this is the one in 4.5 the transfer is almost immediate to the nose so mm yeah so the nose where it’s very very thin its bearing all the weight of the wheel on the right you can see the two designs they’re similar crossing angle but the transfer length is very different they’re both inclined designs but transfer length is very different the top one there’s a Manoir crossing so um you can see their wing rail is quite thin so I’m looking into the different types of designs and looking at what is the optimum wing rail design as well so what is clear is that the wheel transfer all occurred at a crossing nose so as I mentioned wing rail and crossing nose has to work together to optimize the wheel transfer at the crossing in our drawings we have a slope of 32 times crossing angles slow at the nose presumably this is for easy measurement with a straight edge when you go out when you repair the crossings and all that so this hasn’t changed for a long time so the wheel coming to the nose will just go out the slope like that so can you see where the topping ends in the photos so that answers your question for um why is there a gap at the running ban the topping ends there so in fact the wheel is going up a ski slope and

then jumping and causing more damage to their nose and you notice that it all happens on one bearer so this will damage the track support system voiding and all sorts will develop so I did some simulations with a modified wing rail and nose design and then it shows that if you optimize the wing and a nose profile you can actually reduce dynamic forces to a third you’re solving the root cause of the problem whether you have also noticed is that there’s a lot of site batter on crossing nose so I’ve got this photo from they feet this is that Drake Drake’s power state a new crossing his experiencing sight better so um it was apparent that the the check mill wasn’t doing much so if we look at the worst case we’ll check gauge it’s thirteen ninety four point eight mill for normal running so we don’t take the extreme case it’s in 1392 and for passenger stock is thirteen ninety but what are we building our check gauge we are building our design check gauge between thirteen eighty eight to thirteen ninety two so does it cover that will check gauge no so in vertical S&C crossing nose and wheel the crossing nose and wheel can overlap by up to six point five mill so it is possible for the wheel to be striking the side of the nose eating aged common crossings I was asked to go out to this side um like two or three weeks ago we have the new EDH crossing mmm that’s um that’s banging the core banging and voiding is developing so the problem is believed to be false flange damaged so you can see that the running band just changed abruptly from the nose to the leg ends so we put a straight edge on the crossing actually that’s a dip there as you can see from the in a photo which caused sort of at where the running band suddenly changed there’s a dip there so we measured some profiles and then we found out at where the running band changes there’s a dip in the gauged corner I caught the flattening of the gauge corner sudden flattening of the gauge corner so what it means is that as the gauge corner disappear the wheel just falls down so that’s causing the banging on the eth crossing this is using a 1p8 profile it’s not even it doesn’t have folds flange it’s just a warm PA profile you can see that it’s crashing down using a new PA profile it crashed down further it’s 1.5 mil instead so a new wheel would be worse so did some simulations to show that yes indeed you have higher dynamic forces because of this sudden change in the in the running band so I’m still working on common crossings I’m changing the the wing rail and the cross nose design this is from SolidWorks it shows the real going past e from the wing to the nose that’s a dip angle there I need to improve this dip angle optimize it the top graph shows the forces but meanwhile I propose solutions for our cast pressings in the short term specify all crossings with inclined wing rail designs so it will help with support on the wheel and then specify continuous foot flange designs where we can and also I will be specifying new crossing toppings without these 32 times angle slope and also we’re recommending tightening the flange way to 38 mil at a moment it’s 41 mil for check rails at sites with persistent sight better after crossing nose so we’ve much rather prefer the check rail to be worn than the crossing nose and also we have made our first HP premium real check rail so that should

be more wear resistant and also we are specifying manufacturing tolerance for the running table of crossings to make sure that we don’t see any more banging so the medium term we will be looking at the base of the crossing to increase their increase in reducing rigidity and as Ian said we are looking at rail path optimization and also understand in the long term maybe look at different crossing materials so the result from this work is that we reduce we’re aiming to reduce the impact forces a wheel transfer area which would lead to longer crossing life so we reduce voiding and therefore maintenance activities and there will be less well repair than possession times and we can reduce the risk of crack crossings and therefore ESR and associated performance issues this is the last example it’s about check rail Heights on obtuse crossings so what is the issue the the en TS I proposed en TSI addition cess comes up with all sorts of equations it takes into consideration the geometry of the track real bogie and wheel crossing interactions so it specifies the height of the check rail that we need to have so are we compliant what I found is that our race track rails are generally not complying with the new proposed ENT as I standard because if the TI’s TSI standard comes out and we’re not compliant that’s law we have to do something about our crossings that’s too late then hmm so up to use crossing there’s always a necessary gap at the obtuse crossing when the train goes pass it so we need check rails to help guide the wheels so this is the raised check rail of 38 mil so it has more guidance at the back of the wheel if you have normal check rail the guidance length is shorter if we have a bigger wheel the guidance length is longer so we like big wheels we don’t like small wheels on obtuse crossings so um work is ongoing as doing the same thing with their common crossings and to optimize the topping as you can see that’s a big dip angle so as the wheel goes from the knuckle to the nose so we’re trying to optimize this because we have a lot of obtuse crossing failures at the moment so this is what we’re trying to do optimize the crossing so there’s not much dip angle okay so what do we happen on that work 38 male yes our standard says the maximum height on our check will is 38 mil so these drawing shows that most of them are 38 mils so um the ENT si standard the equations if we plot it in terms of charts this is what it looks like so you have a y-axis the minimum wheel diameter and the bottom the type of the diamond angle and the different color lines on the chart are the is the height of the check row so for the ENT si standard the maximum height of the check rail is 60 60 mil but we only have up to 38 milli so if we take an example if we use the 30ml brace check height which is their green one in the middle so if any of our wheel diameters out over there and any of our crossing angles are in there it’s good anything on the right hand side it’s not good okay so I program all the equations into MATLAB and I’ve plotted the graph actually the same graph as you can see for our sin 56 all the lines move towards the left this is because of the narrow gauge 14 32 gauge compared to the same standard the TSI standard which assumes 1535 so all these little changes moves the graph over to the so what is our minimum wheel diameter for for passenger Trek trains I think the minimum is 770 diameter that’s for

the meridians because they have fun in sight bearings so they have smaller wheels but then we have freight trains as well so fka and fla train trains with very very small wheels which is beyond smaller than what the TSI said TSI minimum was 500 but the fla has 461 mill so if we look at passenger stock if we have 38 mill check mill height we can only have that up to a crossing angle of about 6 so anything above I mean wider then I mean sharper than 6 we are non-compliant so these are the fla wagons with the smallest wheel on yeah we have loads of these because I’m the container wagons from Freightliner you can’t even fit a person underneath it so an analysis result is that our last things are complying with the proposed yen TSI for passengers start sort of but our crossings may not be complied with for freight wagons so but that says men mean that they’re not say we haven’t had any derailment obtuse crossings yet so maybe that year the TSI is over stringent you know so we have to liaise with the rssp to influence the standards so current work using modeling I’m investigating the crossing failures at Wessex looking at specific sites like Weybridge Great West Vauxhall and Feltham Junction and I’m also collaborating with our SSB as I said on understanding the implication of a new TSI on raise check rails on our network and I’m also read read sighing probe over an hour 60 switch play to reduce the wear failures in a trailing direction this is known known problem yes the results should be out within the next two weeks mmm so conclusions from my work is that modeling can be powerful for improving systems and component performance however site validation is vital or component confirming the modeling outputs for every work that I’ve used I made sure there’s good coloration correlation between the site measurements and the modeling results because this will give me confidence to make changes to standards processes and drawings example shows tangible benefits in terms of we identified the failure modes and develop solutions for it we check we improve the standards drawings and processes and we improve track designs for easier installation lower canvas and reduce maintenance and therefore lower whole life cost we have good engagement with suppliers to ensure a better quality of products and consistent production processes to improve asset performance and most importantly we have safety by design we make sure we design something that’s safe that we put on track and they will work thank you