Acceptance Criteria for Damaged and Repaired Passive Fire Protection Diane Kerr 1,*, Deborah Wioughby 1, Simon Thurbeck 2 and Stephen Connoy 3 1 Heath and Safety Laboratory, Process Safety Section, UK 2 MMI Engineering Ltd 3 Heath and Safety Executive, Offshore Safety Division, UK * Contact detais: Diane Kerr, Process Safety Section, Heath and Safety Laboratory, BUXTON, SK17 9JN, United Kingdom; e-mai: diane.kerr@hs.gov.uk Crown Copyright 2008. This artice is pubished with the permission of the Controer of HMSO and the Queen s Printer for Scotand Keywords: PFP, Damage, Repair, Cementitious, Intumescent, Acceptance Criteria. Introduction The performance of ageing, weathered, damaged or repaired passive fire protection (PFP) is a major concern, particuary offshore where it is often a safety-critica factor in protecting structures and pant from the effects of severe fires. The two most commony used types of PFP materias are cementitious and intumescent. Cementitious PFP works initiay by hoding the temperature of the substrate to around 100 C unti a the bound water is vaporised and then acts as a passive insuator. Intumescent PFP has an organic base which, when subjected to fire, expands producing a stabe char with good therma insuation properties. Whist the performance of these materias when in good condition is we understood, itte or no information is avaiabe which provides an understanding of their performance in a damaged or repaired state. In the absence of such information, verifying performance standards where PFP is empoyed as a risk-reduction measure cannot be undertaken reiaby. To address this shortfa, a two-phase Joint Industry Project was estabished to deveop data on the performance of damaged PFP and to suggest an inspection methodoogy with acceptance criteria. This paper wi describe speciay produced damaged and repaired PFP test pieces (both cementitious and intumescent) and wi summarise their performance in jet fire resistance tests. The paper wi aso discuss the acceptance criteria deveoped for damaged and repaired PFP. The project s findings are providing important input into a new procedure for identifying when repair or repacement is necessary, and which forms of repair are most effective. The findings aso highight vauabe essons in quaity assurance of the PFP appication process. The work reported here was undertaken in a coaborative project deveoped and managed by MMI Engineering Ltd. The work was sponsored by HSE, BP, Centrica
Energy, Canadian Natura Resources, Marathon, Tota, and She. PFP materias were provided by Internationa Coatings, Cafco Internationa, and PPG Indistries, and specimen coating was undertaken by Saamis and RBG Ltd. Testing was undertaken on behaf of the Sponsors by the HSL. Test Faciity And Instrumentation The tests were carried out using HSL s jet fire resistance test faciity. The faciity is designed to give an indication of how we passive fire protection materias wi perform in a jet fire. It invoves impinging a propane vapour jet fire on a target covered with PFP materia. The test faciities and detais compy with OTI 95 634 and Draft ISO standard ISO/CD 22899-1. The test specimen was boted between an open-fronted stee box (ined with ceramic boarding) and a protective rear chamber. A wa of ightweight buiding bocks was erected around the target assemby to prevent fames impinging on the sides of the target. A photograph showing the instaed assemby is given in Figure 1. The test specimen was instrumented with minera insuated, stainess stee sheathed, 1.5 mm, type K thermocoupes suppied and fitted by HSL. Thermocoupes were instaed in the back surface of the panes by means of dried, interference fit, hoes and were hed in position in the web by bukhead fittings. Measurements were ogged using DT3003 32 channe Data Transation data acquisition boards fitted directy into a PC. A FLIR SC 2000 therma imaging camera was used to measure the temperature distribution over the back of the test specimen. Substrate The substrate for a norma structura stee test consists of an open-fronted stee box, nomina interna dimensions 1500 mm 1500 mm 500 mm, made from 10 mm thick stee. Figure 1. Instaed Assemby (test 10 specimen, before test) 2
It has a 20 mm thick centra web, 250 mm deep, to simuate corner or edge features such as stiffening webs or edges of I beams. The web is made up of two 10 mm thick stee pates, which are sotted before being weded together, to aow the insertion of thermocoupes The box is fanged at the rear to aow attachment (by boting or camping) of a 1500 mm 1500 mm 1000 mm protective chamber when required. In order to be abe to test as many types of damage as possibe, whist keeping the number of jet fire tests to a minimum, it was decided to repace the back of the box with a pane comprising two haf panes and a web. In this way, three different features coud be assessed in each test. A piece of L-section was weded aong the back of one edge of each of the haf panes and hoes dried through these and the back of the web to aow the two haf panes and the web to be boted together (see Figure 2). Figure 2. Haf-Pane and Web Assemby 3
For the repair trias, it was considered that the number of joints shoud be minimised by weding the haf panes and the web together to give a singe pane incorporating the web. A second type of substrate was aso used. This consisted of a simpe pate pane, without a web, fitted to form the back of the box. Coating Test specimens for the damage trias were coated with cementitious or intumescent PFP materias suppied by Cafco Internationa (Mandoite 550), Internationa Coatings (Chartek IV and VII) and PPG Industries (Pitt-Char XP). Specimens were coated with the required thickness of each materia (as specified by the manufacturer) to achieve the foowing protection criteria, when subjected to a jet fire test: Panes at east 60 minutes for a temperature rise of 140 C. Web at east 60 minutes for a temperature rise of 400 C. (temperature rise above initia substrate temperature) Chartek IV is a reativey hard epoxy intumescent PFP materia and Pitt-Char XP is a softer, more fexibe materia. A thickness of 12 mm of the respective intumescent materias was appied to the panes and webs, with reinforcement ocated in the midde third of the coating. The cementitious test specimens were a prepared with a nomina 38 mm deep coating of Mandoite 550. Reinforcement was with pastic coated hexagona wire mesh retained by heica pins in the midde third of the coating. The repair test specimens were coated with a combination of cementitious materia (Mandoite 550) and intumescent materia (Chartek VII or Pitt-Char XP). These particuar intumescent materias were chosen as both can be used with reinforcement that does not require pinning. Coating of the majority of the substrates was witnessed and quaity controed by MMI Ltd. A photograph iustrating coating of a repair test specimen is provided in Figure 3. Test Programme The first test series invoved carrying out jet fire resistance tests on PFP specimens with the foowing types of induced damage: Gouges and nicks; Cracks; Loss of bonding and retention; and Water saturation (cementitious specimen) The second test series was designed to test severa typica in-situ repairs found offshore. In practice, where damage has occurred to a sma area of cementitious materia, it is typicay 4
Figure 3. Coating Appication Butt joint removed and repaced with new intumescent materia. Probems can arise at the interface between the materias, as the two types of PFP work by competey different mechanisms. Three types of repair joint were investigated: Simpe butt joints between the cementitious and intumescent PFP; Two step cementitious/intumescent joints; and A smooth chamfer joint between the materias The induced damage and repair joints are described in more detai in Tabes 1, 2 and 3. The joints are aso iustrated in Figure 4. Resuts and Concusions The most significant resuts and overa concusions from the damage and repair trias are detaied beow. The resuts from this coaborative project wi aso be accessibe, via the HSE website, in due course. Damaged Intumescent Trias Resuts Figures 1 and 5 show the specimen for tria 10 before and after testing. The purpose of this test was to determine the effect of gouging, cracking and disbondment on Chartek IV. Left Pane The eft pane was physicay damaged (gouged), but was bonded to the stee substrate. After the test, sight enargement of the gouges was observed. Thermocoupe temperatures and therma images confirm that the main hot spot was recorded behind the horizonta 5
Tabe 1. Intumescent PFP tests Test number Purpose Anomaies 01 Chartek IV contro specimen 02 Effect of oss of materia, to depth of reinforcement ayer, and beyond on Chartek IV 04 Chartek IV disbonding 10 Effect of gouging, cracking and disbondment on Chartek IV 13 Effect of gouging, cracking and disbondment on Pitt-Char XP None. Panes and web a with same thickness (12 mm) of sound materia and reinforcement Left pane Sound materia with 350 mm 40 mm vertica gouge to depth of reinforcement. Right pane Sound materia with 350 mm 40 mm vertica gouge beyond depth of reinforcement. Web 350 mm ong vertica 45 o nick from edge of web beyond depth of reinforcement. Left pane Disbonded with partiay retained mesh. Right pane Bonded with no mesh Web Disbonded with partiay retained mesh Left pane Sound materia with 200 mm 25 mm horizonta gouge in top haf, 200 mm 12 mm horizonta gouge in bottom haf and 350 mm 12 mm vertica gouge running downwards from centre. A gouges through mesh. Right pane Disbonded materia with 2, 3 and 5 mm cracks to substrate running fu ength. Web Disbonded materia with 5 mm crack to substrate running fu ength. Left pane Sound materia with 200 mm 25 mm horizonta gouge in top haf, 200 mm 12 mm horizonta gouge in bottom haf and 350 mm 12 mm vertica gouge running downwards from centre. A gouges through mesh. Right pane Disbonded materia with 2, 3 and 5 mm cracks to substrate running fu ength. Web Disbonded materia with 5 mm crack to substrate running fu ength. (25 mm 200 mm) gouge above the centre ine. The time unti a 140 C temperature rise was seen behind this gouge was 4 minutes. The highest temperature rise after 60 minutes was 325 C. Right Pane Prior to the test, the right pane was cracked and the coating was deiberatey disbonded from the stee substrate. After the test this pane was much more damaged than the eft 6
Tabe 2. Cementitious PFP tests Test number Purpose Anomaies 05 Contro specimen None. Panes and web a with same thickness (38 mm) of sound materia, and reinforcement 06 Effect of oss of materia, to depth of reinforcement ayer, and beyond Left pane Sound materia with 350 mm 40 mm vertica gouge to depth of reinforcement. Right pane Sound materia with 350 mm 40 mm vertica gouge beyond depth of reinforcement. Web 350 mm vertica nick from edge of web beyond depth of reinforcement. 07 Effect of cracking Left pane Sound materia with 1, 3 and 5 mm cracks running fu ength. Right pane Disbonded materia with 1, 3 and 5 mm cracks running fu ength. Web Disbonded materia with 5 mm crack running fu ength. 08 Effect of moisture content Left pane Sound materia with pane fuy saturated. Right pane Sound materia with pane 50% saturated. Web Sound materia with web fuy saturated. 09 Effect of oss of bonding and oss of retention Left pane Disbonded materia & stee reinforcement fixed around edges of pane. Right pane Bonded materia with no stee reinforcement. Web Disbonded materia and stee reinforcement partiay retained. pane (above), the materia was bowed out from the substrate and cracked. There was considerabe enargement of the induced cracks, particuary the 2 mm crack. The highest temperatures were recorded at the ower ends of the cracks, in the high erosion zone. Therma images confirm that there is a hot area around a three cracks in the high erosion zone. The time unti a 140 C temperature rise was seen behind the cracks was 9 minutes. The highest temperature rise after 60 minutes was 262 C. Web The web was physicay damaged (cracked) and the coating was deiberatey disbonded from the stee substrate. Temperatures rose very rapidy behind the (5 mm) crack on the web. Major damage was caused, with bare stee being exposed for the bottom two thirds of the web. A 400 C temperature rise was achieved after 5 minutes. The maximum temperature rise of 705 C was achieved after 25 minutes, with the temperature rise at 60 minutes being 656 C. These temperatures are consistent with the Chartek IV being stripped off eaving bare stee exposed to the jet fire. 7
Tabe 3. Repair tests Test number PFP Repair 03 Mandoite 550 with hexagona wire mesh reinforcement repaired with Chartek VII with HK-1 gass/carbon fibre reinforcing scrim 11 Mandoite 550 with hexagona wire mesh reinforcement repaired with Chartek VII with HK-1 gass/carbon fibre reinforcing scrim 12 Mandoite 550 with hexagona wire mesh reinforcement repaired with Pitt-Char XP with FM gass fibre reinforcing scrim Simpe butt joint verticay and horizontay in a pane without a web Left pane Stepped cementitious joint across centre ine. Right pane Chamfered intumescent joint across centre ine. Web Stepped cementitious joint on eft side and chamfered intumescent joint on front and right side of centreine. Left pane Stepped cementitious joint across centre ine. Right pane Chamfered intumescent joint across centre ine. Web Stepped cementitious joint on eft side and chamfered intumescent joint on front and right side of centreine. Figure 4. Repair joints 8
Figure 5. Specimen for test 10 (after test) Concusions The main concusions from the damaged intumescent trias were as foows: (a) If there is a gouge in PFP on a pane, a hot spot is formed behind it in a jet fire. If the gouge is down to the substrate, 12 mm and 25 mm wide gouges give a 140 C temperature rise within 3 minutes and a 75 mm wide gouge gives a 400 C temperature rise within the same time. The temperature rise is sower if the gouge eaves PFP attached to the substrate (22 minutes unti 140 C temperature rise on 75 mm gouge with materia remaining. (b) 45 o nicks in PFP on webs ead to bare meta being exposed in a jet fire. Where the nick just exposed the substrate, a 400 C temperature rise was reached in 30 to 36 minutes and where the nick just exposed the mesh, in 46 minutes. The evidence suggests that the erosive forces shear off the char if it expands at right anges to them. (c) A 5 mm crack in disbonded materia has a much more severe effect on materia on an edge feature (web) than on a fat surface (pane). A crack in the PFP on an edge eads to rapid faiure of the PFP with the PFP being stripped off eaving bare meta and giving a temperature rise of 400 C within 4 minutes. This occurs with both Chartek IV (a reativey hard materia) and Pitt-Char XP (a reativey soft materia) and hence it is ikey to occur with any epoxy intumescent materia. Cracks in disbonded materia on fat surfaces sti ead to considerabe damage with a temperature rise of 140 C occurring within 10 minutes, with the Pitt-Char XP being more effected than the Chartek IV. (d) For the materia on the panes, disbondment (with partiay retained reinforcement mesh) and a compete ack of reinforcement appear to have ony a minor effect on fire resistance, with ack of reinforcement having a marginay greater effect (58 minutes for a temperature rise of 140 C compared to greater than 60 minutes). However, as 9
indicated above, if there are faws in the surface of the materia, this can ead to rapid faiure. There appeared to be a sighty greater effect on the disbonded materia with partiay retained reinforcement on the web with a temperature rise of 400 C in 55 minutes rather than the required 60 minutes. (e) Precise specification and quaity contro of intumescent PFP appication (materia thickness, uniformity, reinforcement position) is required in order to achieve the desired eve of fire protection from the PFP. Summary The overa concusions are that the most severe effects are from damage to PFP on edges. Cracks in disbonded materia ead very rapidy to the PFP being stripped off and nicks can give the same effect athough it takes onger. For materia on fat surfaces, gouges or cracks down to the substrate ead to a significance oss in fire resistance, particuary if the materia is aso disbonded. Disbondment or oss of reinforcement aone, (no surface damage) has ony a minor effect on performance provided that the materia is at east partiay attached to the surface (by the reinforcement system if disbonded or by the bonding if the reinforcement system is corroded). Damaged Cementitious Trias Resuts Figures 6 and 7 show the specimen for tria 07 before and after testing. The purpose of this test was to determine the effect of cracking and disbondment on Mandoite 550. Figure 6. Specimen for test 07 (before test) 10
Figure 7. Specimen for test 07 (after test) Left Pane The eft pane was physicay damaged (5, 3 and 1 mm induced cracks), but was bonded to the stee substrate. On competion of the test, the dimensions of the cracks remained unchanged. The highest temperature rise of 166 C after 60 minutes was recorded at the top of the 5 mm crack. Right Pane The right pane was aso physicay damaged (5, 3 and 1 mm induced cracks). In addition, the coating was deiberatey disbonded from the stee substrate. On competion of the test, the dimensions of the cracks remained unchanged but, in this case, the coating bowed out from the substrate in the bottom two-thirds of the pane. The highest temperature rise of 222 C after 60 minutes was recorded at the bottom of the 5 mm crack. The therma image at 60 minutes suggested that the whoe of the right hand pane (with cracked, disbonded materia) was hotter than the eft pane (with cracked, bonded materia). This may account for the coating bowing out from the substrate. Web The web was physicay damaged (cracked) and the coating was deiberatey disbonded from the stee substrate. Initiay the web temperatures around the high erosion positions were higher than the temperatures in the high heat fux positions. Towards the end of the test the temperature at the centre of the web in the high heat fux zone exceeded the others. Bare meta was visibe behind mesh on 95% of front face of the 11
web and on the sides of the web; the PFP materia was (mainy) intact but coming away from the substrate, due to the erosion forces from the jet. If the reinforcement had not been intact, it is ikey that the PFP woud have faen off. A fu-depth crack was observed aong the width of the web at the bottom right hand side. The highest temperature rise after 60 minutes was 473 C behind the midde of the 5 mm crack on the web. Concusions The main concusions from the damaged cementitious trias were as foows: (a) 350 mm 40 mm gouges on fat surfaces resuted in sighty higher temperatures if the gouge was through the retention mesh rather than down to the mesh athough the damaged panes sti both easiy met the 140 C temperature rise in greater than 60 minute fire resistance requirement. The 350 mm ong 45 nick in the web gave much higher temperatures than an undamaged specimen but sti met the 400 C fire resistance criterion. Gouges and nicks of the size used woud not resut in faiure, athough the safety margin is ower for a nick on an edge feature. (b) For the specimens with 1, 3 and 5 mm cracks, cracks in sound materia were ess serious than those in disbonded materia. None of the cracks increased in size, probaby because the retention mesh was intact. It is ikey that the PFP on the web woud have faen off if the retention mesh had been corroded by water ingress. Cracks in sound materia reduced the fire resistance time by 5 minutes whereas cracks in disbonded materia reduced the time by 15 to 17 minutes. (c) 100% saturation and 50% saturation of sound materia made no significant difference to the fire resistance. There were no signs of water saturation causing spaing. (d) Disbondment with fu or partia retention appeared to have itte effect on the PFP performance if there were no openings to aow jet the jet fire to get between the materia and the substrate. Disbondment with cracking, especiay on an edge feature, eads to significant oss of fire resistance. (e) If the materia is fuy bonded to a fat surface and there are no jet fire ingress points, ack of retention appears to have no effect. However, oss of retention from the bottom of a gouge has a sight effect and from a nick in an edge feature has a significant effect. Summary The argest temperature differences and consequent eariest times for water to be driven off a occur on a damaged web suggesting that damage to edge features is again most critica. The maximum substrate-heating rate after the water was driven off was 5.7 C min -1. Athough the test with water-saturated specimens had to be prematurey terminated, there was no evidence to suggest that these specimens woud not have met their ratings. The ony specimens not to meet their fire resistance specification were those with cracks. 12
Repair Trias Resuts Figures 8, 9 and 10 show the specimen for tria 11 before and after testing. The purpose of this test was to determine the effect of chamferred and stepped repair joints. Left Pane The eft pane incorporated a stepped repair joint. On this joint, fuy expanded firm char was formed that was intact except for a crack near the edge of the pane on the eft hand side where a sma gap was visibe where the Chartek had come away from Mandoite (see Figure 9). Right Pane The right pane incorporated a chamfered repair joint. At the interface between the Mandoite 550 and chamfered Chartek VII (see Figure 10), the Chartek had not come away from the Mandoite. However, there was a sight crack aong the base of the Mandoite and aong the Chartek. On the right side of the joint the char was more eroded and the mesh was exposed over a sma area. Web The eft hand side of the web incorporated a stepped repair joint. On the front of the eft (stepped) side of the web, there was some damage to the Mandoite 550 and char had been eroded away near this zone. Figure 8. Specimen for test 11 (before test) 13
Figure 9. Left pane after test (11) Figure 10. Right pane after test (11) 14
Figure 11. Therma images (at 60 minutes) of repair joints The right hand side of the web had a chamfered (Chartek VII) repair joint. After the test, the char on this side of the web was not firm and the mesh was exposed in the jet impact zone. Overa, the temperature pots and therma images indicated that, athough both types of joint appeared acceptabe, the chamfered joint performed better than the stepped joint. Therma images comparing a three repairs are iustrated in Figure 11. Concusions The foowing concusions were made in regard to the three types of repair assessed: (a) Butt joints between sound cementitious materia (34 mm thick) and new intumescent materia (12 mm thick) are the weakest of the three types of joint assessed. When the jet impacts on the thinner coating of intumescent materia, it is directed across the surface unti it hits the edge formed by the thicker coating of cementitious materia, where it is then directed away from the surface. The combination of high erosive forces and high heat fux appears to resut in the char eroding away faster in this configuration. (b) Stepping the sound cementitious materia and overcoating this with new intumescent materia gives an adequate join but there are sti erosive effects and, because the thickness of intumescent materia is not much greater than norma, the repair is not as good as chamfering the intumescent materia. (c) Chamfering the intumescent materia with an extra oop of reinforcement appeared to be the best form of repair. It worked we with both a reativey hard intumescent materia and with a reativey soft one. In addition, it worked as a repair on an edge feature. Acceptance Criteria for Damaged PFP As reported by Thurbeck (2006), current Offshore Safety Case submissions assume that PFP materia is fit-for-purpose, performs as specified, and therefore performance standards 15
Tabe 4. Acceptance criteria for anomaies in epoxy intumescent PFP Anomay Criteria Suggested action Loss of topcoat Cracking Disbonded Loss/remova of materia on structura steework Loss of topcoat is ACCEPTABLE. Fuy bonded, fuy or partiay reinforced, materia which contains cracks of ess than 2 mm width is ACCEPTABLE. Materia which is disbonded or has damaged reinforcement, and has cracks of any width or depth is UNACCEPTABLE. Disbonded materia of an area of ess than 1 square metre, and which shows no signs of cracking or surface anomaies, is ACCEPTABLE. Disbonded materia of an area of greater than 1 square metre, or a disbonded area of ess than 1 square metre which has surface anomaies, is UNACCEPTABLE. Loss of bonded, reinforced materia over an area which is ess than the acceptabe area, and to any depth of remaining materia which may incude damage to the reinforcement, is ACCEPTABLE. Action shoud comprise remova of oose topcoat materia to a cean and sound materia, a check that adequate thickness remains, and then retopcoating with a recommended topcoat. A check shoud aso be made to identify the presence of corrosion at the substrate surface. The crack shoud be seaed and made watertight to prevent water ingress into the materia which may each out active agents associated with the intumescent process. Action shoud comprise remova of the materia down to the substrate, and back to soundy bonded materia, and the impementation of the Manufacturers recommended repair. No repair is necessary but the anomay shoud be monitored. Repair shoud comprise remova of the materia down to the substrate, and back to soundy bonded materia, and the impementation of the Manufacturers recommended repair. Action shoud comprise remova of any oose materia and the appication of an appropriate topcoat system. 16
are met. Whist this may be true, itte or no information is avaiabe which provides an understanding of PFP performance in a damaged or repaired state. Verification of the performance standards coud not, therefore, be reiaby undertaken. Required fire performance is generay specified in terms of time and temperature. In simpe terms, the fire performance criteria specify surviva times and are generay associated with ensuring an appropriate eve of integrity is maintained to prevent an event from escaating, or for ensuring that personne can be evacuated safey. The fire performance requirements for a critica item shoud be detaied in the performance standards, aong with the avaiabiity and reiabiity of the eement, and the verification tasks to ensure that the performance is ensured. The criteria are described by defining: Whether the protected item is a barrier or a oadbearing eement; The type of fire/heat fux oading to which it is subjected; The critica temperature which shoud not be exceeded to ensure integrity and/or insuation requirements are met; and The time over which the temperature shoud remain beow this critica temperature. The data from the damage and repair trias described above, aong with practica considerations, have been reviewed and a set of acceptance criteria produced, incuding a decision fowchart. This can be used by asset integrity personne to assess the acceptabiity, or otherwise, of observed anomaies in PFP coatings and to determine whether the performance standard of a particuar Safety Critica Eement is maintained. Exampes of the type of anomaies that (without corrective action) coud be expected to ead to a faiure (or reduced performance) incude: Disbonded materia of an area of greater than 1 m 2. Materia which is disbonded or has damaged reinforcement, and has cracks of any width or depth. Loss of materia, regardess of area, which is down to the substrate or has insufficient thickness to provide the required fire resistance performance. Acceptance criteria for cementitious PFP and epoxy intumescent PFP materia containing anomaies wi be accessibe, via the HSE website, in due course. An extract from the criteria, incuding suggested action where anomaies are observed, is shown in Tabe 4. References Heath & Safety Executive, 1996, Offshore Technoogy Report OTI 95 634, Jet-fire resistance test of PFP materias. ISO standard ISO/CD 22899-Parts 1 & 2, (draft). 17