A Egieer's Casebook Articles reproduced from ICI Safety Newsletters 101-171 (July 1977 October 1983)
No 1 Overhaul, Repair, Recoditio - What Do We Mea? Item 101/1 i this Newsletter referred to a leak which developed at the body joit of a HP valve leadig ultimately to a fire. Itesive ivestigatio afterwards revealed that the specified close fit betwee the spigot ad recess of the trapped joit, dimesioed o the detail drawig of the valve, had bee lost. The diametral clearace should ot have exceeded 0.002 ich; i fact it varied from 0.021 ich to 0.029 ich. Clearaces of this magitude are too great for full cotaimet of the 1 /32 ich thick CAF gasket material, a requiremet of a trapped joit. The valves cocered were set out for overhaul ad repair to a workshop which had udertake such work for decades. However, very few of these special valves ow remai i service ad most of the me who were familiar with them have retired, particularly the supervisors ad ispectors, who, i the absece of writte overhaul ad recoditioig procedures, applied their ow stadards based o experiece. Close ispectio showed up other udesirable features i recoditioed valves ad brought home the eed to specify what is meat by overhaul ad repair or recoditio as required. I geeral it will be ecessary to do more tha quote the origial Stadard or Drawig No sice these rarely idicate allowable weardow toleraces. It will be ecessary to agree with the repairer precisely what is to be doe, whe, ad how, so that recoditioed items are truly equivalet to ew oes. E H Frak
No 2 Ruig Policy For Spare ad Multiple Uits Most equipmet is subject to wear ad tear whilst it is i service ad if it is left ruig log eough it will evetually fail. Where more tha oe item of equipmet is i service, if all have a approximately equal exposure to the service coditios, the they may be expected to fail after roughly the same umber of ruig hours. You might say that this is statig the obvious; quite so, ad quality cotrol over maufacture, precise desig ad reproducibility ted to esure that there is a miimum scatter i the time to failure. We have the followig recet examples. Four axial flow compressor rotors (3 workig, 1 spare) are regularly rotated roud the three workig machies. As a result they have similar ruig hours, i this case about 6 years. Blade crackig i oe was rapidly followed by failure i aother ad examiatio of the remaiig two also revealed cracked blades. I aother case a fleet of four doze rail tak cars, of idetical desig ad costructio, all had, omially, the same umber of ruig hours. They all developed cracks i the barrels at about the same time as a result of which there was a major repair requiremet ad a shortage of cars whilst the repairs were carried out. Situatios like this ca be avoided by esurig that oe of the operatig uits is deliberately ahead of the others i terms of ruig hours, thereby givig it a opportuity to reveal ay built-i weakess. I the case of multiple uits phased ispectio ca also help. Istead of ispectig the whole populatio every say, 5 years a umber should be doe each year, thereby providig a opportuity of early warig of time-depedet failure by corrosio, fatigue, wear, etc. E H Frak
No 3 Bellows I 1971 there was a survey i the Divisio o the use of expasio bellows ad a report, available oly i ICI, was issued (No PC.200, 812/A). The report, which was commissioed after two expesive major plat shut dows caused by o-lie failure of metallic expasio bellows uits, commeted that The itroductio of a expasio bellows ito a plat i most cases gives rise to a weak lik i a system which otherwise is costructed to a uiform stadard of mechaical stregth. The report aalysed the cause of 53 reported bellows failures; 20 of these were due to stress corrosio crackig by caustic or chlorides ad 17 were due to a iitial costructio or istallatio fault. Two recommedatios were made to reduce these causes. The first was that bellows should preferably be made from Icolloy 825 to elimiate stress corrosio crackig by chlorides, though ot by caustic. The secod was that bellows should be istalled i the field i a way similar to that used for critical items of equipmet. Sice 1971 the umber of bellows i ew plats has bee sharply reduced ad i a few istaces existig uits were replaced by pipig systems with iheret flexibility. The 1972 revisio of the ICI Pressure Vessel Code (ICI Egieerig Code ad Regulatios, Group B, Volume 1.4, available outside ICI from the Royal Society for the Prevetio of Accidets, Queesway, Birmigham B4 6BS) required the registratio of selected expasio bellows i kow erosive/corrosive service or where a high hazard would be created by failure. Registratio required regular ispectio. Ufortuately failures cotiue, though at a much reduced level followig actios take ad geeral publicity, due to icomplete cotrol. I a recet icidet a large uit, operatig at 600 psi, i the ilet lie to a steam turbie blew apart oly a few hours after it had bee istalled. The uit, which was brad ew, relied o split rigs bolted roud its covolutios to support them ad equalise expasio betwee the multiple corrugatios. The accidet ivestigatio foud that some of the split rigs ad oe of the ed skirt rigs were slack with gaps of up to 9/16 ich betwee the butts of the half rigs. The gaps, otably the oe roud the first rig, removed most of the support from the first corrugatio allowig the uit to squirm ad through that to precipitate total failure by rupture where the Icolloy 825 membrae was welded to the stub eds. The equiry could ot establish why some of the split rig bolts were slack or if ayoe had deliberately slacked them, for example, to have a look at the corrugatio udereath. It did establish that o precommissioig ispectio took place ad there was every chace that had oe bee carried out the slack rigs would have bee corrected. Registered bellows uits must be treated like pressure vessels ad subjected to a thorough examiatio by a equipmet ispector before eterig service ad at the prescribed routie iterval thereafter. E H Frak
No 4 Pipework Our plats cotai a lot of pipework most of which attracts little or o attetio ad oly a few pipes are subject to regular examiatio. Pipe wall thickess must be such as to cotai the stress due to pressure otherwise failure will result. Stress arisig from other causes, for example thermal expasio or dead weight loadig, may exceed the yield stress without a failure occurrig. The pipig takes up a slightly differet cofiguratio to hold the stress at the yield value. Provided this sort of thig does ot take place too frequetly, i which case a fatigue failure may result, or the extra dead weight is so severe as to cause gross displacemet by bedig, thigs will be alright. The smaller sizes of pipe usually have a hadsome degree of over-desig built ito them so that they tolerate the ormal wear ad tear of beig o a productio uit. For sizes 8 iches ad above, ad for higher pressures, desig is more closely related to allowable stress levels. Above 12 iches or so pipig is ormally tailored to the particular duty ad purpose desiged for it, usig calculated wall thickess. Cosequetly it is importat that it is istalled as the desiger iteded ad operated withi the desig limits of pressure, temperature etc.; there is o extra arbitrary factor of safety built i as i the smaller lie sizes. I 1972 a 22 ich IP steam mai developed a crack after some years of service. This might be described as a fairly severe duty carryig superheated steam at 250 psig ad 365 C, certaily a applicatio where the desig was based o calculated wall thickess ad adequate flexibility to accommodate the thermal expasio withi the allowable stress rage. After the failure, which was i the form of a circumferetial crack o the bottom of the pipe ad may ot have bee due solely to overstress, the lie ru was checked agaist the desiger s drawig. The followig fidigs are quoted from the report o the icidet. Sprig support H1 had its sprigs fully compressed. Sprig support H2 was show o the drawig but ot fitted. Sprig support H3 was i positio but was ot attached to the pipe. Sprig support H4 was i positio but the uts o the ed of the support rods were slack. Whe the stresses i the lie were computed i the as foud coditio the maximum was 60 000 lbf/i2, more tha twice the maximum allowable. Failure was ot surprisig. It is importat that the larger diameter lies o a plat are give a pre-commissioig check for compliace with the desiger s drawigs. Hagers, guides ad achors are ot optioal extras but essetial features, particularly o the larger pipelies. E H Frak
No 5 SPM SPM - Safe Productio Methods, or perhaps Scietific Priciples of Maiteace, both equally commedable. However, i this case, SPM stads for Shock Pulse Measuremet. What is it? It is a simple meas of determiig, o lie, the coditio of a rollig elemet bearig. The priciple of operatio is based o the trasmissio of shock waves through the bearig housig, the waves emaatig from miute damage o the cotactig surfaces. Damage to the races, cages or rollig elemets themselves arises from pittig or debris produced after a period of properly lubricated operatio, by uder-lubricated operatio, ifiltratio of dirt particles etc. A trasducer attached to the machie by a special sesor stud screwed ito a bearig housig, or clamped to a bearig housig usig a modified pair of plier grips, is used to pick up the shock wave. The trasducer is tued mechaically ad electrically to have a resoat frequecy of 32 khz. The shock waves set up a dampeed oscillatio i the trasducer at the resoat frequecy. The peak amplitude of this oscillatio is related to the magitude of the shock wave ad is cosequetly a measure of the coditio of the bearig. The trasducer sigals are processed i a shock pulse meter which gives a scale readig of shock value. A oogram is provided to relate this value to the bearig coditio. Alteratively the rage readigs idicatig good coditio, damage likely ad bearig damaged are well-established. All rollig cotact bearigs cotai mior imperfectios which result i a backgroud shock value readig at a low level for a ew healthy bearig. SPM is used to moitor this level ad to detect the oset of icipiet damage. There has, of course, to be some icrease i shock wave productio by way of tramplig more debris before SPM ca alert a potetial problem ad it might, by the, be said that the writig is o the wall. The equipmet is ot therefore truly prevetive. It does however have cosiderable merit i detectig deterioratio at a early stage so that replacemet work ca be plaed before eforced outage ad possible cosequetial damage esues. The maufacturers claim that the read-out ca be iterpreted to idicate defects or malfuctios other tha deterioratio of bearigs by displayig the output o a cathode ray tube or graphically. It should ot however be cosidered as a substitute for covetioal vibratio moitorig equipmet. SPM equipmet has bee used by Orgaics Divisio for some years ad a useful paper describig the equipmet together with some actual case histories has bee writte by two of their Egieers. E. H. Frak
No 6 Oil Lubricated Bearigs Oil lubricated bearigs cosume oil whilst i use through spillage, vaporisatio, leakage from reservoir etc., thus creatig a eed to repleish the reservoir where bearigs are desiged for cotiuous lubricatio. A simple ad coveiet way of doig this o small reservoirs, such as those o most pumps ad fa bearigs, is to use a costat level oiler. A umber of differet proprietary oilers are i use o the Divisio s Works ad all use the same priciple which is that of a bottle type chicke feeder. A iverted bottle has a tube secured to the cetre of the bottle cap so that oil from the bottle ca ru dow the tube. The tube, which may be from ¼ ich to ½ ich i diameter ad ¾ ich to 3 iches log, depedig o the desig of the oiler, has its ed chamfered at 45 degrees. The iverted bottle is mouted, usually adjacet to the reservoir, at a level such that the top of the chamfer o the feed tube is at the oil level to be maitaied i the reservoir. As the oil level falls the top of the chamfered feed tube is exposed ad air eters the tube risig up ito the bottle. A equal amout of oil is released, the process cotiuig util the pre-set oil level is restored whe the chamfered tube is sealed off ad oil feedig ceases. I some cases the chamfered oil feed tube dips directly ito the level i the reservoir or sees the same level through a partially full side brach. I others the surface fed by the bottle is etirely separate from that i the reservoir beig oe leg of a U tube which coects the bottle to the reservoir. I such cases it is ecessary to esure that the surface fed by the bottle is ope to the atmosphere ad a vet hole(s) is provided for this purpose. A recet pump bearig failure may have bee due to the vet becomig blocked i a rather uusual way. The oil bottle was of Deco maufacture ad has a split brass sleeve surroudig the separate oil ad air feed pipes which protrude from the cap. This sleeve is soft soldered ito the cap ad two vet holes, each about 1/16 ich diameter, are drilled through the cap ad sleeve after they have bee sweated together. The sleeve is quite a tight push fit i the body of the adaptor ad it is likely that operators twist the bottle whe isertig or withdrawig it. The sweated joit betwee the cap ad sleeve had sheared allowig oe to rotate with respect to the other thus blakig off the two small vet holes. The holes ca be easily checked with a paper clip ad whe this was doe o all Deco bottles o the Works cocered oe further case of a rotated sleeve was foud ad aother case where the vet holes had bee paited over. Some other potetial causes of malfuctio which may apply are:- Adjustig sleeve set too low o Deco bottles arraged for side feedig, chamfered tube set at wrog height, bottle fittig or adaptor loose where screwed ito reservoir (should be positively locked agaist rotatio), leaks i screwed fittigs, faulty bottle to cap joits, occluded bottle prevetig sight of true oil level etc. A careful check roud all costat oiler istallatios might prevet uecessary failures. A specificatio based o the better features of existig oilers ad desiged to remove error producig features is beig prepared ad might be icorporated ito the ICI Bearig Lubricatio Specificatio E1. E H Frak
No 7 Throttle Bushes The Divisio s Specificatio MAC 0304 for cetrifugal pumps ad API Recommed Practice 610, Cetrifugal Pumps for Geeral Refiery Purposes, defie the requiremets for the majority of cetrifugal pumps used i the Divisio s Works. Where pumps are fitted with a sigle mechaical seal, ad most are, there is a requiremet for a o-sparkig throttle bush to be fitted ito the seal ed plate. The diametral clearace betwee the iside diameter of this bush ad the shaft passig through it is specified to ot exceed 0.025 is. Bushes may be made of brass, sitered broze, PTFE etc. The purpose of the throttle bush is to restrict the amout of the pumped fluid which is discharged to atmosphere i a ucotrolled way should there be major leakage from, or total disruptio of, the mechaical seal. I a way it acts as a crude auxiliary sealig device ad limits the discharge of product to eable the pump to be shut dow, prevets the productio of a large flammable cloud of vapour if liquefied flammable gas (LFG) or hot flashig liquids are discharged ad helps direct spillage alog the drai coectio from the space outboard of the seal. The throttle bush should ot be cofused with the throat bush (or eck bush) which is at the iboard ed of the seal chamber ad serves to limit the clearace betwee the shaft ad pump casig ear the back of the impeller. Uder ormal circumstaces the maximum diametral clearace of 0.025 is betwee the throttle bush ad shaft should more tha cater for shaft displacemets arisig from impeller imbalace, bearig wear etc. Oe might suppose that there should be o cotact, o wear ad therefore o eed to check or replace the throttle bush. Last year a check was carried out o oe Works i which the diametral throttle bush to shaft clearace was measured o eleve pumps o LFG duty. O oly two was the clearace withi the specified 0.025 is, some exceeded 0.100 is the greatest beig 0.180 is. It had ot bee this Works practice to check throttle bush clearaces durig pump overhauls. This is ow beig doe ad clearaces brought ito lie with the desig figure. The source of hazard approach to area classificatio assures that the clearace is ot greater the 0.025 is. The maiteace of this is a ecessary part of cotiued safe operatio where pumps hadle LFG, flashig liquids ad other materials which ca give rise to sigificat clouds of flammable vapour. E H Frak
No 8 Routie Test Pressure Occasioally it may be ecessary to apply a hydrostatic test to equipmet which has bee i service for a period of time as a meas of demostratig that it is fit for further service. I geeral such tests are of limited value cofirmig oly that, at the time of test, the equipmet is satisfactory ad givig o idicatio of future life. Tests which measure metal thickess, allow visual examiatio, avoid cotamiatio through the use of test water etc. are therefore the preferred methods. Nevertheless it will be ecessary to hydrotest where other forms of examiatio are impracticable, major repairs or modificatios have bee carried out with a eed to apply a proof test, a secodhad vessel is ivolved etc. Whe a hydraulic test is to be applied oe must first check to see if the weight of water has bee allowed for i the foudatio/support structure desig ad that due allowace is made for the effect of static head. The oe has to establish the value of the test pressure itself. I the past it was always fairly safe to assume that all equipmet was desiged o the basis of it beig full of water ad that, the test pressure was 15 times the desig pressure. Nowadays thigs have chaged ad it is essetial to cosider each hydraulic test as a separate exercise. Except for simple routie testig operatios it is appropriate to cosult the Desig Authority to establish the viability of a hydraulic test ad the magitude of the test pressure. It is a matter of policy i Petrochemicals Divisio to desig foudatios ad support structures for ay vessel o the basis that it is full of water. It is ot the practice however to desig the shell thickess of larger vessels, for example, distillatio colums over 100 ft. tall, o the basis of hydrostatic pressure at the base whe full of water plus the maximum desig operatig pressure. Vessels are built to oe or other of a variety of desig codes. The less striget of these, measured through factors of safety, costructioal stadard, evaluatio of stress, utilisatio of material properties, for example, BS 1500, use a test factor of 1.5 based o the as-built thickess at the weakest part. Advaced desig codes ivariably use proof test factors which are less tha 1.5, allow for ay loss of material stregth at temperature if i hot service, ad deduct ay corrosio allowace whe calculatig the test pressure. The factor for BS 1515 is 1.3 ad for BS 5500 1.25. Equipmet registratio cards ofte quote routie test pressure whilst maufacturers drawigs quote iitial test pressure. It is advisable ot to use these figures as the basis for ay hydraulic test applied to equipmet which has bee i service but rather to ask the Desig Authority for guidace o the pressure to be applied. E H Frak
No 9 Nuts ad Bolts The frustratio which we have experieced for years with fidig the right size of spaer to fit a particular hexago ut is becomig icreasigly apparet i fidig a ut to fit a particular bolt. I the case of spaer sizes it is readily apparet that if the spaer is a slack fit you eed to rig the chages betwee the three mai types, amely, Whitworth, A/F ad metric, to fid oe which does fit properly or, as alas may do, resort to mole grips or a Stillso wrech. Ultimately oe is usually successful i gettig the ut tighteed dow by oe meas or aother. But what has bee achieved if the ut ad bolt are mismatched? Most of the uts ad bolts i use o our plats are required to produce a miimum bolt load, for example, to yield a gasket i a pipe joit, to carry pressure ad dyamic loads i reciprocatig machiery, high stregth frictio grip bolts i structural coectios etc. A few uts serve merely to hold the bolt i place where the bolt is i shear. It is therefore importat that ot oly is the ut tight o the bolt but also that the ut ad bolt are properly matched, each of correct diameter, pitch of thread ad thread form so that whe correctly tighteed the desig bolt load is produced. Three screw thread systems cover most of the uts ad bolts used o our plats from the ¼ ich size upwards ad cosiderable error potetial exists for them to become iterchaged. The three thread systems are the Whitworth system (BS 84), traditioally used i the U.K. ad based o the roud topped 55o Whitworth thread form, the Uified ISO (ich) system (BS 1580) based o the America 60o flat top thread form ad the ISO Metric system (BS 4190) which also uses a 60 thread form. ICI took a lead i this Coutry i the late 1940 s i pioeerig the itroductio of the Uified thread series ito what at that time was a almost exclusively Whitworth field. The widespread itroductio ad adoptio of the Petroleum Equipmet Egieerig stadards for petrochemical plat based o the America ASA series, otably for pipig, accelerated the ivestmet i Uified boltig i the late 1950s ad 1960s, largely through the use of stud bolts to BS 1750. More recetly the move towards metricatio, ow well established i the electrical ad civil ad buildig idustries, ad importatio of more goods from withi the EEC is icreasig the umber of uts ad bolts made to the ISO Metric system. This will ultimately become the oly system to be used. Iterchageability betwee the three mai thread systems is illustrated o the accompayig chart (BSF ad UNF igored) which shows that thread egagemet ca vary from full egagemet of a loose ature of icorrect uts from aother thread system, through misleadig apparetly correct egagemet, to a tight fit of perhaps oly two or three threads which could tempt the user to apply force suspectig a burr, dirt or other cause of tightess. Proper segregatio i stores helps keep the problem uder cotrol together with a awareess of error potetial o the part of all users. Markig systems ca help where maufacturers ad/or users have applied these. Whitworth uts ad bolts carry o idetificatio, Uified bolts have a shallow circular recess o the head with uts stamped o oe flat with circles (000) or a circular groove of semi-circular sectio i the o-bearig face, ISO Metric bolts have the symbols ISOM or M stamped or embossed o the head with uts stamped with a M o oe flat. ICI Stadard Sheet 080200 gives all the details. E H Frak
No 10 Rollig Cotact Bearigs Rollig cotact bearigs, i.e. ball races, roller races, agular cotact thrust bearigs etc. will ot last forever. Costat slight deformatio at the poit of cotact betwee the rollig elemets ad the races betwee which they ru, essetial to give a cotact area to carry the radial ad/or axial load, gradually fatigues eve the best material. Fatigue leads to the detachmet of small pieces of metal from oe or other surface. These pieces of metal are the trampled o by the rollig elemets which further icreases the loadig o material which is already tired ad fairly rapid failure results. The life of a rollig bearig may be calculated, life represetig the period of uimpaired performace of the bearig whe properly mouted, adequately lubricated, protected from foreig matter ad ot subjected to extreme operatig coditios. If a reliability is chose as the probability that a bearig will attai a certai specified life, or exceed it, the a basic ratig life ca be defied as the life associated with 90% reliability. Calculatio methods for dyamic load ratigs ad ratig life are give i BS 5512 Part 1 1977 which is a copy of ISO 281/1-1977. L10 (formerly B10) is the basic ratig life i millio revolutios. For a give speed of rotatio this ca be easily coverted ito hours. For example, a L10 life of 40000 hrs, which is the preferred selectio for pump bearigs i our Egieerig Specificatios, relates to 3600 millio revolutios at 1500 rpm. 90% of idetical bearigs operatig uder the same coditios should reach this milestoe before the first evidece of fatigue develops. I practice premature failure usually takes place due to operatio uder o-ideal coditios, dirt i the oil, vibratio, periods at rest givig local effects (Briellig) etc. Rollig bearigs which have ru for 16-25000 hours (2-3 years) should be cosidered for replacemet eve though they may still appear, ad feel, i excellet coditio. Their expectatio to survive a further 8-16000 hrs (1-2 years), which is a fairly commo life betwee overhaul of much equipmet, is pretty low.
No 11 Have You The Right Spares ad Istructios? O restartig a turbo-compressor after a major overhaul it soo became apparet that some iteral fault was allowig gas from oe part of the compressor to leak ito aother part so cotamiatig the outlet. The machie was desiged so that this should ot happe. However, as we seem to fid all too ofte, the origial system did ot work too well. It was modified 8 years ago ad had bee quite satisfactory from the up to the overhaul. Plat staff at Egieer ad Supervisory level had chaged i the iterveig years ad those i charge at the recet overhaul were uaware of the previous history ad modificatios which had bee made to the compressor. I advace of the overhaul, which was carried out by a team of me from the maufacturer s service departmet, those spare parts which it was thought might be eeded were draw out of stores from the plat spare gear refereces. Amogst those fitted was oe part, a shaft labyrith with a special modificatio for this machie. This part had udergoe a desig chage as part of the 1970 modificatios ad though a spare to the ew desig had bee ordered, ufortuately (you have guessed it) the old spare had ot bee throw away ad was iadvertetly fitted. As a result the plat had to be shut dow to allow the compressor to be opeed up agai ad the fault rectified; a expesive error. Somewhere alog the lie a obsolete capital spare had ot bee writte off or had a maiteace istructio bee prepared to detail the special requiremets of this otherwise apparetly stadard machie, It is easy at the time of chage to thik I kow all about this, why should I write a istructio ad We had better keep that old desig as a stadby, we could always modify it i a emergecy. Time passes, memories fade, staff chage..! How are your spares ad istructios, all ship-shape for tomorrow s geeratio? E H Frak
No 12 Examiatio of Statutory Vessels: Hydrotest? Some egieers are still of the opiio that Statutory equipmet (steam boilers, steam receivers ad air receivers) requires a hydraulic test as part of its Statutory Examiatio. This is evideced by a recet icidet ivolvig a steam receiver i this Divisio which was damaged whilst o test ad a equiry from aother Divisio suggestig that perhaps hydrotestig was ot ecessary. The ote o the back of Form 58, Note B, ad o form 59, Note C, merely require the occupier to provide facilities for a thorough examiatio to take place. The metio of hydraulic test, steam trial, or other meas of testig are idicative of the sort of work which the competet perso coductig the thorough examiatio may deem to be ecessary. What is ecessary will be partly determied by the costructio of the receiver, access for ispectio, service coditios etc. Sectio 35 of the Factories Act 1961 does ot call for a hydraulic test every 26 moths; it says every steam receiver ad its fittigs shall be thoroughly examied by a competet perso, so far as the costructio of the receiver permits, at least oce i every 26 moths. The decisio as to what costitutes a thorough examiatio is i the hads of the competet perso. He has to examie the receiver i such a way as to satisfy himself that o part is weakeed i ay way such that it is o loger capable of safe operatio at the workig pressure. I past years, with riveted receivers, before the use of radiography to determie the state of welded structures, o crack detectio methods, ultrasoic thickess testig etc., competet persos had little alterative, where thorough visual examiatio ad calliper thickess measurig was ot possible, but to determie adequacy through a proof hydraulic test. Today this is ot so; ispectio aids are readily available for use both o ad off lie. As a result, i most cases, cotiued fitess for purpose ca be determied by meas other tha by pressure testig. Ideed iformatio obtaied via thickess testig is much more meaigful i determiig whether metal is beig lost, so that future life ca be forecast, tha is the successful passig of a pressure test which says oly that, at the time of the test, there was sufficiet metal. Hydraulic proof testig to levels of 1.5 or 1.3 times the desig pressure, depedig o the desig code used, ca result i the material yieldig at certai places where stresses are cocetrated or there are residual stresses from pipig, weld cotractio etc. Proof testig also raises the level of stress at defects such as weld udercuts, slag iclusios, blow holes, cracks, ad may spread these, but ot to the poit of ultimate failure, so that after test the vessel is less adequate tha it was beforehad. Uecessary proof testig is to be discouraged. Testig up to the set pressure of the safety valve, usually the desig pressure, is, of course, quite acceptable ad ofte useful to check for tube ad flage leaks. E H Frak
No 13 Ispectio of Protective Devices Sometimes the obvious is overlooked durig ispectios or tests carried out to check fuctio, perhaps because the tests are ot sufficietly demadig. Two recet, true cases ca be used to illustrate these poits. Small sized burstig discs, particularly those of 1 ich diameter ad less, which are desiged to fit ito uio type holders rather tha carriers, are ofte supplied with a gasket glued oto each side of the disc. This facilitates hadlig ad reduces the chace of accidetal damage to which small discs are particularly proe, a typical thickess beig 0.005 ich. A ivestigatio ito the failure of a disc to operate whe it should have doe so discovered that the maufacturer had iadvertetly supplied two discs which, ested oe o top of the other, appeared as though they were oe. The bodig-o of the gaskets preveted the edges beig checked to verify that oly oe burstig disc was there. A check by the supplier, who i this case acted as a stockist, revealed three more double discs i his stock. The check was doe by weighig each disc icludig its boded-o gaskets. Double discs weigh 0.6 g more tha the 1.1 to 1.2 g of a sigle disc with gaskets. Most discs are supplied as sigles idividually boxed to avoid damage; however some of the heavier discs may be foud i a stack. It is worth a look at your istructios to see that there is a built-i check to esure that double discs are ot fitted. I aother case a tak roof was ruptured by overpressure oly 3 moths after the atmospheric vet had bee checked ad apparetly proved clear. After the accidet the vet pipe was foud to be choked with rust scale at a bed beeath a vertical leg which led upwards to the atmosphere. The tak was oly desiged for 2½ iches water gauge pressure uder the roof. The test to prove the vet lie clear was to discoect it from the tak top, isert a air hose ito the the ope ed of the vet lie ad tur o the air. Provided that the air disappeared up the vet lie ad did ot blow back uduly the vet was assumed clear. Such a test is completely oquatitative; quite a large amout of air could pass through a lie obstructed to the equivalet of a 1 ich diameter hole. It was ot possible to see through the vet lie; it is rarely possible to do so. Vet lies ad relief valve discharge lies are a itegral part of protective devices which prevet equipmet beig overpressured. Their ispectio forms part of the protective device examiatio ad the ature ad extet of the examiatio is at the discretio of the Competet Perso. Judgemet has to be exercised. Part of that judgemet must be based o the likely effect of back pressure o the protected equipmet. Clearly taks desiged for oly a few iches water gauge pressure caot tolerate vets choked with pipe scale, perhaps ot eve with bird s ests! Vets from such taks must be subjected to checks desiged to show that they are completely uobstructed. E H Frak
No 14 LP Steam Mai Expasio Bellows ad Achor Failure Much of the equipmet o the Divisio s plats is registered ad ispected i accordace with the ICI Code B1.4 Registratio ad Periodic Ispectio of Pressure Vessels. As such it receives regular attetio by the Divisio s Equipmet Ispectors ad Plat Egieers. Some low pressure pipig carryig o-hazardous fluids is ot registered. However that is ot to say that it is immue from failure. This photograph is of a achor with axial expasio bellows uits o either side istalled i a LP steam mai. Failure was iitiated through operatig the mai at a temperature i excess of that for which it was desiged. It is likely that the damage took place may moths, if ot years, ago though it was oly oticed withi the last six moths. We should all be observat whilst carryig out the daily roud ad commo tasks. E H Frak
No 15 Slip-O Flages: A Uusual Failure Slip-o flages are cosiderably cheaper tha weld eck flages ad are i widespread use o the less rigorous pipelie duties. As the ame implies, the flage is slipped over the outside diameter of the pipe util the ed of the pipe is short of the flage face by a amout equal to the pipe wall thickess. Fillet welds are the used to secure the pipe to the flage, the mai stregth beig i the back weld to the hub. There is cosiderable diametral clearace betwee the iside diameter of the flage ad the outside diameter of the pipe, 1/16 ich for 2 ich.b. pipe icreasig to ¼ ich for 24 ich.b. pipe. This clearace aids aligmet ad fit up i the field. Pipig desig codes BS.3351 ad ASA B31.3 both cotai the followig cautioary ote: The use of slip-o flages should be avoided where may large temperature cycles are expected, particularly if the flage is ot isulated. It will be oted that o quatitative guidace is give. What is may? 100, 1000 or more? What is large? 200, 300, 400 C? A plat usig class 300 stailess steel slip-o flages ad thi wall stailess steel pipe ra ito trouble after about eight year s life which represeted about 1800 temperature cycles. Flage welds started to crack ad ultimately major repairs were ecessary to all the slip-o flages i hot cyclic service. A few simple calculatios demostrated that very high stresses were iduced i the welds betwee the pipe ad flage by differetial thermal expasio as the system warmed up. The thi wall pipe used had a low heat capacity ad was i direct cotact with the process fluid. The flage, o the other had, had a large capacity ad was relatively slow to respod to risig fluid temperature. The use of stailess steel, which has a coefficiet of liear expasio about half as much agai as carbo steel, aggravated the problem. I this particular case a differetial temperature well i excess of 100 C was measured betwee the pipe wall ad flage hub as the system warmed up to its operatig temperature of 450 C. The correspodig thermal strai gave rise to a stress level of over 1500 N/mm2 (more tha 100 tos/sq ich) i the welds. This stress, though applied for oly a few miutes, resulted i total rupture, by crackig, after 1800 cycles. Guidace give i desig codes whereby the allowable desig stress ca be used without reductio up to a cyclic life of 7000 applies oly where the actual level of stress is withi that allowable by the code for the particular material used. E H Frak
No 16 Spiral Woud Gaskets: ½ ich to 1½ iches omial size Metallic spiral woud (SW) gaskets to BS 3381 : 1973 are i widespread use throughout the Divisio ad i pressure ratigs of Class 300 ad above have largely replaced compressed asbestos fibre. May egieers may be uaware of the BS sice the use of brad ames such as Metaflex ad Flexitallic have become syoymous with SW gaskets. I the size rage ½ ich to 1½ iches NS there is a problem i as much as the stadard gaskets to BS 3381 are suitable oly for weldig eck flages. With slip-o or screwed flages, where the bore of the flage is larger, a stadard gasket protrudes beyod the flage faces ito the pipe itself eve whe correctly cetred by the cetrig rig withi the bolt circle. At the iside diameter of a SW gasket are three or so turs of the metal strip which are woud o to each other ad spot welded together before further widig ad itroductio of the filler strip. These turs form the foudatio of the gasket ad are iteded to be gripped betwee the flages whe the joit is bolted up. The applicatio of the flage bolt load results i a iward radial force o the SW gasket. Whe usig a stadard gasket betwee a pair of slip-o flages i which the ier turs are ot compressed this force ca break the metal strip ad allow the sealig elemet to uwid ito the bore. The exposed turs are also beig buffeted by the pipelie fluid. Both BS 3381 ad maufacturers catalogues cotai warigs that the stadard gaskets should be used oly with weldig-eck flages i the sizes ½ ich to 1½ iches. However, sice o other gaskets are listed i BS 3351 or catalogues, stadard gaskets ted to be the ormal supply. Ufortuately most of the flages used o our plats i the ½ ich to 1½ iches NS are slip-o flages; cosequetly may SW gaskets are i service i these sizes uder o-ideal coditios. Maufacturers ca supply special SW gaskets which are dimesioally suitable for slip-o or screwed flages i which the whole of the sealig elemet is trapped betwee the raised faces. Maufacturers idetify these specials by stampig SO o the outer guide (cetrig) rig. Stadard gaskets carry o flage idetificatio, oly size, class ad material. Nomially at least, SW gaskets supplied as specials for slip-o flages should be suitable for use with weldig eck flages sice the bolt loads are exactly the same. Maufacturers, however, seem reluctat to recommed their use, perhaps because they do ot coform to the dimesios of stadard gaskets i BS 3381. A attempt is beig made through BSI to have a suitable rage of SW gaskets i sizes ½ ich to 1½ iches for slip-o flages detailed i BS 3381. Meawhile tradesme should be aware of this problem. Your SW gasket stocks should predomiatly be specials stamped SO or wholly this type, thereby elimiatig the chace of icorrect selectio. E H Frak
No 17 Registratio ad Examiatio of Pipework This week has oce agai see a eforced shut dow of oe of our major sigle stream plats whe a leak developed o a blaked brach. A corrosive by-product of the mai reactio had codesed i the cooler brach ad over may years of operatio had, at oe place, goe right through the pipe wall thickess. Examiatio of braches i other places where coditios were similar revealed similar corrosio ad a shutdow of several days is ecessary for repairs. There was a history of corrosio i the feed/product heat exchager ad attempts had bee made to check the adjacet pipig. Nothig had bee foud probably because the mai pipe wall was above the critical temperature for depositio of the corrodet. Whilst the ispectio of pressure vessels is well ordered ad regulated through the rules i the ICI Code Group B1.4, pipework ofte fails to get the attetio it deserves. Whe the Code was revised i 1972 a ew category, B8, was itroduced to apply to selected pipework i kow erosive/corrosive or high hazard service. There has bee a reluctace to idetify may systems i this category ad few pipig systems are registered o our plats. This may have bee due to lack of guidace o registratio ad ispectio methods. Code B 1.4 is curretly repritig ad icludes additioal guidace o the problem of pipig. Sectio 3 has bee expaded to iclude a separate sectio o the registratio ad classificatio of pipig systems. This idetifies the criteria for registratio which is required whe:(i) a system or part thereof is kow or suspected to deteriorate from corrosio, erosio, fatigue or other factor AND (ii) is i such service, ad locatio, that failure would give rise to a uacceptable situatio. The resposibility for registratio lies with each Works; whe takig actio the ispectio orgaisatio ad materials egieers will be able to offer advice ad assistace. Registratio creates a ispectio task ad sice the meas ad techiques available for this are i some cases differet from those traditioally used for pressure vessels guidace is required. A ew Sectio 14 has bee writte for this purpose ad is i the revised editio ow at the priters. By the time this Safety Newsletter reaches you copies of the revised Code will have bee issued. If you have ot reviewed pipework o your plat by ow ad registered those systems as appropriate you should be doig so. E H Frak
No 18 The Use of Pressure Vessel Desig Codes Most of the pressure vessels which are used i the petrochemical or chemical idustries are desiged ad built to a iteratioal Code. A few special purpose vessels, such as air receivers, may be made to other British Stadards Pressure vessel codes i regular use are:bs 1500 Part 1 Fusio Welded Pressure Vessels for geeral purposes - Carbo ad low alloy steels BS 1515 Part 1 Fusio Welded Pressure Vessels for use i chemical, petroleum ad allied idustries - Carbo ad ferritic alloy steels BS 1515 Part 2 Fusio Welded Pressure Vessels for use i chemical, petroleum ad allied idustries - Austeitic Stailess Steels BS 5500 Ufired fusio welded pressure vessels ASME VIII Divisio 1 Rules for costructio of pressure vessels ASME VIII Divisio 2 Rules for costructio of pressure vessels - alterative rules BS 1515 ad ASME VIII Divisio 2 are sometimes referred to as higher desig stress codes requirig more precise desig procedures, certified material properties etc. The recet BS 5500 code uses a similar approach. It should be oted that i May of this year BS 1500 ad BS 1515 were withdraw followig the issue of BS 5500 i 1976. BS 5500 is therefore the oly curret British Pressure Vessel Code. It is more comprehesive tha the codes which it has replaced ad icludes desig rules for tubesheets. Pressure vessels costructed to BS 1500 ad 1515 were, of course, quite adequate vessels ad copies of these codes will eed to be retaied for referece whe repairs, modificatios or upratig/deratig queries arise. New vessels should be desiged ad costructed to BS 5500. There is a tedecy to use pressure vessel codes to determie shell ad head thickess, compesatio required at braches, stresses at supports etc., i.e., to use the code formulae ad allowable stresses, whilst disregardig may of the other code requiremets relatig to quality assurace durig costructio. This is probably at its worst with the use of ASME VIII Divisio 1 (very few vessels are ordered to Divisio 2). There are few maufacturers i this coutry who are audited ad approved by the America Society of Mechaical Egieers after which, followig strict adherece to the verificatio durig maufacture procedure, the maufacturer is permitted to stamp the fiished vessel with the ASME U stamp. May vessels are made allegedly to ASME VIII Divisio 1 the scatligs of which may have bee calculated from that Code but quality cotrol durig maufacture has bee that established by the maufacturer. Verificatio of material properties, welder competece, iterpretatio of radiographs ad the like is ofte left to the maufacturer with some spot checkig perhaps o behalf of the purchaser. Such vessels are ot really built to the ASME Code. Aother commo error is to use steel to British Stadards for which the ASME Code does ot give allowable stresses. Equivalet values are used. The icomplete ad hybrid applicatio ad use of pressure vessel desig codes is techically icorrect. Where it caot be avoided steps must be take to esure that the ed product is at least as good as would have bee the case had the full code requiremets bee applied. These steps should be recorded, i writig, ad form part of the pressure vessel history. I future, particularly for vessels i hazardous plats, it will be appropriate to esure that all vessels have met all aspects of code requiremets ad carry a formal certificate to this effect. E H Frak
No 19 Burstig Discs ad Holders A spectacular icidet i the Divisio, which occurred about two years ago whe a burstig disc assembly did ot prevet a reactio vessel beig seriously overpressured (see Newsletter 98/1), resulted i a examiatio of the desig, istallatio ad checkig of such devices. The device failed to operate because it had bee assembled the wrog way roud. The reactor had to be heated durig oe part of the batch process ad, as ofte happes with such equipmet, it was possible o coolig for a vacuum to be created i the vessel. A vacuum or reverse pressure support was cosequetly specified to support the disc agaist iwards collapse uder these circumstaces. Clearly vacuum supports should be placed o the ilet pressure side betwee the vessel brach ad the disc. They are riddled with holes so that they do ot prevet the pressure i the vessel beig applied over the full area of the burstig disc. Ufortuately, i this case, the vacuum support had bee assembled o the exhaust or vetig side of the burstig disc. As the reactor pressure rose, due to a ruaway reactio, the burstig disc membrae was preveted from rupturig at the desig pressure sice it was beig supported by the vacuum support behid it. I fact it ever ruptured eve though the pressure rose to perhaps early twice the desig pressure. Fortuately the excessive pressure started leaks at the bolted cover ad flaged joits ad these relieved the pressure thereby prevetig possible disruptio of the reactor. Curret practice has bee to rely o traiig tradesme i the correct assembly of all devices, to display exploded assembly sketches i workshops, to have a ewly fitted device checked by a supervisor etc. Ufortuately, o close examiatio, there is cosiderable opportuity for error i fittig, particularly followig the itroductio of reverse bucklig discs which must be fitted the opposite way roud to covetioal domed discs. May maufacturers desigs are difficult to check for correct assembly after they have bee bolted ito the pipelie. Ideed with some desigs a ispectio after assembly is impossible. As a result of the ivestigatio the Divisio has prepared Egieerig Specificatio P1 0601, Covetioal Domed Burstig Discs ad Holders. This requires istallatios to comply with certai features aimed at the elimiatio of much of the error potetial iheret i existig istallatios. Assemblies are ot iheretly safe by desig; however, all vital features ca be checked by ispectio after the disc has bee istalled i the plat. This check is a essetial part of safe operatio. The mai features covered i the ew specificatio which is beig applied to all future projects are: All discs shall be tagged by the maufacturer usig a tag which is rivetted or welded to the disc ad exteds beyod the outside diameter of the flages whe the, disc ad holder is istalled. Tag to carry at least:- maufacturer, omial bore, material, order No., tag No., burst pressure. at.. Tags to be marked vet side at the extremity of the tag ad made from corrosio resistat materials. Vacuum supports to be expedable type ad permaetly attached to the disc by the maufacturer. Holders ad discs to be of the shaped seat (coical or spherical) type wherever possible. Holders to carry a flow leged with arrow idelibly marked to idicate the correct directio of flow, As existig stocks of discs come up for reewal adoptio of the ew specificatio should be cosidered as a meas of reducig or elimiatig icorrect assembly. Note the specificatio does ot cover reverse domed ad graphite discs or covetioal discs less tha 1 ich diameter i plug or uio type holders. E H Frak
No 20 Shakedow From time to time ew words may come ito regular use to describe thigs which, whilst kow about, may ot have bee specifically labelled. Shakedow is such a case. A fabricated structure, for example a pressure vessel, will after fabricatio have locked ito it various stresses. These arise from plate formig, pressig, local doggig to get plates ito aligmet prior to weldig, burig or otherwise cuttig holes i the shell, weld cotractio effects etc. Some of these may be partially removed by a stress relief heat treatmet which ca be applied to particularly critical parts or to the whole vessel. Eve with fully stress relieved vessels some residual stress will remai ad this may be sigificat if the vessel has udergoe a proof hydraulic test after heat treatmet. Whe the vessel eters service ad is loaded by the operatig coditios of pressure, temperature, weight of cotets, pipig reactios etc., the level of stress at ay poit will be determied by the combiatio of the applied loads ad momets. The level of stress will ot be uiform throughout the structure eve though the desiger may have tried his best to esure this. The stress will be cocetrated or itesified i local areas, typically i the back of flages, i the viciity of opeigs, at chages of sectio, i way of supports, attachmets ad over saddles etc., ad by residual effects metioed above. O the first applicatio of the maximum loadig at ay place i the stressed structure, which might be durig the iitial hydrotest or, i the case of vessels operatig at temperature is ofte at plat start-up, the level of stress ca be so high that the material i a limited local area yields ad behaves plastically. Plastic deformatio causes a chage i shape with cosequet re-distributio ad dimiutio of the level of stress. If whe the loadigs are removed ad subsequetly reapplied to a level ot higher tha the origial loadig the behaviour of the material is totally elastic, the the vessel is said to have shake dow. Repeated applicatio of a ever icreasig load or excessive cyclig close to the boudary betwee elastic ad plastic behaviour ca result i premature failure. Whe usig ay of the higher desig stress Codes such as BS 5500 or ASME VIII Div. 2, the formula therei allows the calculatio of the stress level at all sigificat poits. It is recogised that shakedow has a part to play i the redistributio of stress i local areas of high stress. BS 5500, Appedix 6, Clause 2.6 gives a method for radial ozzles i spherical shells whereby the shakedow factor for iteral pressure loadig, for radial thrust ad exteral momet, may be calculated. A proof hydrostatic test will result i a certai amout of shakedow ad ca be positively beeficial. Repeated applicatio, i doubtful cases, which results i yieldig ad plastic deformatio ca result i icremetal collapse or fracture uder test. E H Frak