[
Existing tests supporting the standard for reactive coatings for structural steel do not require columns to be loaded – as long as comparable loaded beams have been tested. Norman Macdonald examines whether loaded columns would produce more realistic results.
To ensure buildings are reasonably safe for their occupants and property is protected in the event of a fire, most of them are divided into fire compartments and the loadbearing walls, floors, beams and columns are required to have a suitable level of fire resistance. In steel-framed buildings it is vital that the structural steel columns and beams are fire protected in order to provide adequate levels of fire resistance.
The use of structural steel to form the framework of a building can be an attractive alternative to concrete structures, and UK steel consumption is currently estimated at about 1.3m tonnes a year. As a rough estimate, about 60-70% of all the steel that requires fire protection in the UK is coated in fire protection paint, with about 10-20% protected using boards and about 10% with cementitious sprays. Fire protection paints are often referred to as reactive coatings.
The market for reactive coatings has grown rapidly over the last few years and is now worth about GB pound 20m a year in the UK. In other European countries the proportion of steel sections protected using reactive coatings is quite different. For example, boards are most common in Germany and cementitious spray in France. However, it is expected that the proportion of structural steelwork protected with reactive coatings throughout Europe will grow. For life safety and property protection, it is important that the performance of any of these fire protection systems is properly assessed.
The existing European test standard for applied protection to steel members is DD ENV 13381-4:20021 and is referred to in the guidance documents supporting building regulations throughout the UK. This is a Draft for Development and although it is currently under review to make it a full test standard, it is still provisional until the test laboratories and manufacturers gain more experience in testing fire protection systems to its provisions. It is proposed to divide the document into two – EN 13381-41 for boards and sprays and EN 13381-82 for reactive coatings.
This article outlines the findings of a recent BRE Trust project which was designed to contribute to the development of the standard for reactive coatings. The aim of the project was to confirm whether or not a loaded column should be included in the test standard.
Reactive coatings
First of all, what are the reactive coatings that can be applied to loadbearing structural steel beams and columns to prevent them collapsing prematurely in fire conditions? They are coatings that expand (intumesce) to many times their original thickness to form an insulating carbon char when exposed to heat. Most coatings are solvent borne or water borne. The use of these systems has increased noticeably in recent years due to their low thickness, attractive appearance and low cost. About 75% of the coatings used to protect structural steel in the UK are applied off site, thus reducing the construction time on site.
Work over many years has clearly demonstrated that the fire performance of a structural steel section partially depends on the surface area of steel exposed to the fire, and the mass or volume of steel that is being heated. The greater the surface area, the faster the steel will heat up; the heavier the steel section, the longer it will take to heat up. These two factors are combined in what is known as the section factor, as shown in figure 1.
Ideally, to assess the fire performance of a reactive coating system, one would fire test every available size of beam and column under load, protected with a range of coating thicknesses. However, the number of different steel section sizes makes such an exercise impractical. The practice over many years in most countries, therefore, has been to test a limited number of coated loaded beams and columns, and compare their results with those of tests on a range of coated unloaded sections. There may be a variation between the results of the loaded and unloaded sections due to the effect of the deformation and expansion of the larger loaded steel section, and the possible shrinkage or detachment of the charred coating. If there is any variation between the loaded and unloaded sections, the results from the unloaded sections are corrected and then analysed as shown below by applying a correction factor.
This same approach is used in the draft European standard EN 13381-82. However, the European standard does not require loaded vertical columns to be tested if a loaded horizontal beam has been tested. It has been proposed by some manufacturers and test bodies that any variation in the performance of the coating on a loaded section compared to that on an unloaded section would be revealed by testing the loaded beam, so it is unnecessary to also test a loaded column.
The draft European standard does include an unloaded column that is 2m tall and therefore taller than the other unloaded sections at 1m height each. This taller column is included in an effort to detect any variations in performance due to the length of the section. It is also proposed in the draft that this unloaded section has a few extra thermocouples near the top of the column, to detect areas of steel that may be exposed due to the ‘slumping’ of the charred coating.
There is concern, however, that this procedure may not detect any other areas of steel that have become exposed directly to the fire due to shrinkage or detachment of the charred coating. Exposed areas of steel that are located some distance from the thermocouples attached to the steel may not detect the higher temperatures in that area. Such areas may be more critical than similar areas on a horizontal beam, particularly if they occur near the ends of the beam.
Project objectives
In order to determine the effectiveness of the test requirements for thin film reactive coatings for the fire protection of loadbearing steel columns, the objectives of this BRE Trust project were as follows:
– To determine whether or not it is adequate to only test unloaded vertical columns, if a loaded horizontal beam has already been tested
– To compare results of loaded sections with unloaded sections
– To determine how close thermocouples have to be located to ‘hot patches’ in the steel.
This latter objective was felt useful to demonstrate the possibility of the unloaded columns not detecting hot spots. Such hot spots can be caused by the char of the reactive coating cracking and shrinking back, or simply falling off the steel, in localised areas. These hot spots may cause premature collapse of loaded vertical columns. The same extent of hot spots on a loaded horizontal beam is unlikely to cause such a rapid collapse of the section, as the load may be shared by the floor slab above.
Test programme
A reactive coating system available in the market was selected for the test programme. The following sections were tested:
– Unloaded 2m column
– Unloaded 1m column fixed to roof of furnace
– Unloaded 1m column on floor of furnace
– Loaded 3m column
– Loaded 3m column – small area of char intentionally knocked off during the test
– All sections 300mm x 300mm x 8mm square hollow sections
– All sections protected with nominal 1.8mm reactive coating which was expected to achieve about 60 minutes’ fire resistance based on previous test results
All of the steel sections tested were coated with a nominal 1.8mm thickness of reactive coating by the coating manufacturer, under the supervision of the BRE Fire Resistance test laboratory. The coating system was expected to achieve about 60 minutes’ fire resistance based on previous test results. The steel temperatures of each of the sections were monitored using the method and locations stipulated in the draft of the test standard current at the time of test, as shown in table 1 (Note: a= Hole made in char at mid-height of column and b= Hole cleared of char that has formed over the hole made at 30 minutes). Additional thermocouples were secured on one of the loaded columns at the location where it was intended to remove some of the charred coating during the test as described above.
Results
The coating in all test specimens behaved in a very similar manner. On each section the coating expanded to form a char about 100mm thick. Unexpectedly, the first loaded 3m column failed loadbearing capacity after 32 minutes.
During the tests on all the sections it could be seen that the coating on each section shrank away from at least two corners of the column. The drawing also shows the area of the second loaded column where the char was deliberately removed after 30 minutes at mid height on one face. Thermocouple positions are shown by an ‘x’. The test results are summarised in the table below.
Analysis
If a system performs well and provides an even level of protection over the entire surface of the column, one would normally expect a hollow steel column to fail the loadbearing capacity performance criterion of the standard when the mean steel temperature is about 5200C. However, if there are limited areas where the protection system falls off or shrinks, then the column may well collapse prematurely due to a localised buckling hot spot. At this time the mean steel temperature will be lower than 5200C as demonstrated in the two loaded column tests in this project.
As explained earlier, the draft European test standard prEN 13381-82 requires that the results from the unloaded sections are used to carry out the thermal analysis, but only after they are corrected by comparing them with the 2m unloaded column. The draft method uses the ‘characteristic temperature’ of each section to determine the correction factor. The characteristic temperature is the mean temperature plus the maximum temperature divided by two. This method of calculating the characteristic temperature should be influenced by any hot spots on the section, but the extent of this influence tends to be variable.
During this project, it was considered important to calculate the correction factor for the case where the loaded 3m column was used instead of the unloaded 2m column. These calculations resulted in the correction factor based on the loaded column being better than the factor determined when using the 2m unloaded column result. This initial finding in isolation would indicate that using the unloaded 2m column result is at least as onerous as using the result from a loaded column. However, it is important that the maximum permissible fire resistance period in the analysis of any results recognises the loadbearing capacity of the loaded section. By only applying the correction factor to the results on the unloaded 1m column, as required by the current draft, the analysis concludes that the applied coating thickness would ensure the column has a fire resistance of 47 minutes. During the testing programme, however, the loaded column was unable to bear the applied load (loadbearing capacity) after only 32 minutes, resulting in a fire resistance of 32 minutes when the loadbearing capacity of a 3m column was measured.
Some buildings require long fire resistance periods, in excess of one hour. In these cases it is possible that the discrepancy between the results obtained with and without loaded column data could be much greater. It is clear that the existing draft method is failing to detect this discrepancy. So in terms of both life safety and property protection issues, it is vital that a loaded section is incorporated in the test programme to impose an upper limit on the calculated fire resistance periods in terms of loadbearing capacity and predicted coating thicknesses.
It was also clear from the results of this work and a closer examination of the individual measured temperatures, that the thermocouples located at the standard positions were not detecting the exposed areas of steel at the corners of the sections, or at the area of steel that was deliberately exposed after 32 minutes on the second loaded column. The results are summarised in table 1 and show that only 75mm away from the centre of an exposed area of steel the measured steel temperature was about 650C lower than in the centre of the exposed area when the mean steel temperature was about 5200C.
Conclusions
The results from this work have shown that a loaded vertical column may be unable to bear the applied load at the times predicted using the test results of only unloaded sections. The temperature of hot spots may not be detected by the thermocouples located at the positions required in the draft European test standard prEN 13381-82, which means that a test on a loaded column would be important.
These tests have only been carried out on rectangular hollow sections but they do suggest that similar results could be obtained from circular hollow sections and I-sections. These findings confirm previous testing experience which also demonstrated that hollow sections can sometimes perform worse than similar I-sections because of their shape. It would therefore seem sensible to ensure that the European test standard includes the requirement to test loaded vertical column sections of the different profiles as appropriate.
The results from this work only relate to tests on one particular intumescent protection system for structural steelwork. However, they clearly show that the results obtained using the draft European standard have over-predicted the loadbearing capacity of the sections, when based on testing of loaded beams and unloaded column sections only. The over-prediction of performance by 15 minutes is significant for a product designed to achieve 60 minutes’ fire resistance. These results suggest that the adoption of prEN 13381-82 without the inclusion of a loaded column test could have serious implications for
the structural stability of buildings in fire.
Norman MacDonald is with BRE Global. The author gratefully acknowledges the support of the BRE Trust in funding this project.
References
1. DD ENV 13381-4:2002. Test methods for determining the contribution to the fire resistance of structural members. Applied protection to steel members. BSI, London, 2002.
2. prEN 13381-8. Test methods for determining the contribution to the fire resistance of structural members. Part 8: Applied reactive protection to steel members. Draft dated March 2007.
[
Existing tests supporting the standard for reactive coatings for structural steel do not require columns to be loaded – as long as comparable loaded beams have been tested. Norman Macdonald examines whether loaded columns would produce more realistic results.
To ensure buildings are reasonably safe for their occupants and property is protected in the event of a fire, most of them are divided into fire compartments and the loadbearing walls, floors, beams and columns are required to have a suitable level of fire resistance. In steel-framed buildings it is vital that the structural steel columns and beams are fire protected in order to provide adequate levels of fire resistance.
The use of structural steel to form the framework of a building can be an attractive alternative to concrete structures, and UK steel consumption is currently estimated at about 1.3m tonnes a year. As a rough estimate, about 60-70% of all the steel that requires fire protection in the UK is coated in fire protection paint, with about 10-20% protected using boards and about 10% with cementitious sprays. Fire protection paints are often referred to as reactive coatings.
The market for reactive coatings has grown rapidly over the last few years and is now worth about £20m a year in the UK. In other European countries the proportion of steel sections protected using reactive coatings is quite different. For example, boards are most common in Germany and cementitious spray in France. However, it is expected that the proportion of structural steelwork protected with reactive coatings throughout Europe will grow. For life safety and property protection, it is important that the performance of any of these fire protection systems is properly assessed.
The existing European test standard for applied protection to steel members is DD ENV 13381-4:20021 and is referred to in the guidance documents supporting building regulations throughout the UK. This is a Draft for Development and although it is currently under review to make it a full test standard, it is still provisional until the test laboratories and manufacturers gain more experience in testing fire protection systems to its provisions. It is proposed to divide the document into two – EN 13381-41 for boards and sprays and EN 13381-82 for reactive coatings.
This article outlines the findings of a recent BRE Trust project which was designed to contribute to the development of the standard for reactive coatings. The aim of the project was to confirm whether or not a loaded column should be included in the test standard.
Reactive coatings
First of all, what are the reactive coatings that can be applied to loadbearing structural steel beams and columns to prevent them collapsing prematurely in fire conditions? They are coatings that expand (intumesce) to many times their original thickness to form an insulating carbon char when exposed to heat. Most coatings are solvent borne or water borne. The use of these systems has increased noticeably in recent years due to their low thickness, attractive appearance and low cost. About 75% of the coatings used to protect structural steel in the UK are applied off site, thus reducing the construction time on site.
Work over many years has clearly demonstrated that the fire performance of a structural steel section partially depends on the surface area of steel exposed to the fire, and the mass or volume of steel that is being heated. The greater the surface area, the faster the steel will heat up; the heavier the steel section, the longer it will take to heat up. These two factors are combined in what is known as the section factor, as shown in figure 1.
Ideally, to assess the fire performance of a reactive coating system, one would fire test every available size of beam and column under load, protected with a range of coating thicknesses. However, the number of different steel section sizes makes such an exercise impractical. The practice over many years in most countries, therefore, has been to test a limited number of coated loaded beams and columns, and compare their results with those of tests on a range of coated unloaded sections. There may be a variation between the results of the loaded and unloaded sections due to the effect of the deformation and expansion of the larger loaded steel section, and the possible shrinkage or detachment of the charred coating. If there is any variation between the loaded and unloaded sections, the results from the unloaded sections are corrected and then analysed as shown below by applying a correction factor.
This same approach is used in the draft European standard EN 13381-82. However, the European standard does not require loaded vertical columns to be tested if a loaded horizontal beam has been tested. It has been proposed by some manufacturers and test bodies that any variation in the performance of the coating on a loaded section compared to that on an unloaded section would be revealed by testing the loaded beam, so it is unnecessary to also test a loaded column.
The draft European standard does include an unloaded column that is 2m tall and therefore taller than the other unloaded sections at 1m height each. This taller column is included in an effort to detect any variations in performance due to the length of the section. It is also proposed in the draft that this unloaded section has a few extra thermocouples near the top of the column, to detect areas of steel that may be exposed due to the ‘slumping’ of the charred coating.
There is concern, however, that this procedure may not detect any other areas of steel that have become exposed directly to the fire due to shrinkage or detachment of the charred coating. Exposed areas of steel that are located some distance from the thermocouples attached to the steel may not detect the higher temperatures in that area. Such areas may be more critical than similar areas on a horizontal beam, particularly if they occur near the ends of the beam.
Project objectives
In order to determine the effectiveness of the test requirements for thin film reactive coatings for the fire protection of loadbearing steel columns, the objectives of this BRE Trust project were as follows:
– To determine whether or not it is adequate to only test unloaded vertical columns, if a loaded horizontal beam has already been tested
– To compare results of loaded sections with unloaded sections
– To determine how close thermocouples have to be located to ‘hot patches’ in the steel.
This latter objective was felt useful to demonstrate the possibility of the unloaded columns not detecting hot spots. Such hot spots can be caused by the char of the reactive coating cracking and shrinking back, or simply falling off the steel, in localised areas. These hot spots may cause premature collapse of loaded vertical columns. The same extent of hot spots on a loaded horizontal beam is unlikely to cause such a rapid collapse of the section, as the load may be shared by the floor slab above.
Test programme
A reactive coating system available in the market was selected for the test programme. The following sections were tested:
– Unloaded 2m column
– Unloaded 1m column fixed to roof of furnace
– Unloaded 1m column on floor of furnace
– Loaded 3m column
– Loaded 3m column – small area of char intentionally knocked off during the test
– All sections 300mm x 300mm x 8mm square hollow sections
– All sections protected with nominal 1.8mm reactive coating which was expected to achieve about 60 minutes’ fire resistance based on previous test results
All of the steel sections tested were coated with a nominal 1.8mm thickness of reactive coating by the coating manufacturer, under the supervision of the BRE Fire Resistance test laboratory. The coating system was expected to achieve about 60 minutes’ fire resistance based on previous test results. The steel temperatures of each of the sections were monitored using the method and locations stipulated in the draft of the test standard current at the time of test, as shown in table 1 (Note: a= Hole made in char at mid-height of column and b= Hole cleared of char that has formed over the hole made at 30 minutes). Additional thermocouples were secured on one of the loaded columns at the location where it was intended to remove some of the charred coating during the test as described above.
Results
The coating in all test specimens behaved in a very similar manner. On each section the coating expanded to form a char about 100mm thick. Unexpectedly, the first loaded 3m column failed loadbearing capacity after 32 minutes.
During the tests on all the sections it could be seen that the coating on each section shrank away from at least two corners of the column. The drawing also shows the area of the second loaded column where the char was deliberately removed after 30 minutes at mid height on one face. Thermocouple positions are shown by an ‘x’. The test results are summarised in the table below.
Analysis
If a system performs well and provides an even level of protection over the entire surface of the column, one would normally expect a hollow steel column to fail the loadbearing capacity performance criterion of the standard when the mean steel temperature is about 5200C. However, if there are limited areas where the protection system falls off or shrinks, then the column may well collapse prematurely due to a localised buckling hot spot. At this time the mean steel temperature will be lower than 5200C as demonstrated in the two loaded column tests in this project.
As explained earlier, the draft European test standard prEN 13381-82 requires that the results from the unloaded sections are used to carry out the thermal analysis, but only after they are corrected by comparing them with the 2m unloaded column. The draft method uses the ‘characteristic temperature’ of each section to determine the correction factor. The characteristic temperature is the mean temperature plus the maximum temperature divided by two. This method of calculating the characteristic temperature should be influenced by any hot spots on the section, but the extent of this influence tends to be variable.
During this project, it was considered important to calculate the correction factor for the case where the loaded 3m column was used instead of the unloaded 2m column. These calculations resulted in the correction factor based on the loaded column being better than the factor determined when using the 2m unloaded column result. This initial finding in isolation would indicate that using the unloaded 2m column result is at least as onerous as using the result from a loaded column. However, it is important that the maximum permissible fire resistance period in the analysis of any results recognises the loadbearing capacity of the loaded section. By only applying the correction factor to the results on the unloaded 1m column, as required by the current draft, the analysis concludes that the applied coating thickness would ensure the column has a fire resistance of 47 minutes. During the testing programme, however, the loaded column was unable to bear the applied load (loadbearing capacity) after only 32 minutes, resulting in a fire resistance of 32 minutes when the loadbearing capacity of a 3m column was measured.
Some buildings require long fire resistance periods, in excess of one hour. In these cases it is possible that the discrepancy between the results obtained with and without loaded column data could be much greater. It is clear that the existing draft method is failing to detect this discrepancy. So in terms of both life safety and property protection issues, it is vital that a loaded section is incorporated in the test programme to impose an upper limit on the calculated fire resistance periods in terms of loadbearing capacity and predicted coating thicknesses.
It was also clear from the results of this work and a closer examination of the individual measured temperatures, that the thermocouples located at the standard positions were not detecting the exposed areas of steel at the corners of the sections, or at the area of steel that was deliberately exposed after 32 minutes on the second loaded column. The results are summarised in table 1 and show that only 75mm away from the centre of an exposed area of steel the measured steel temperature was about 650C lower than in the centre of the exposed area when the mean steel temperature was about 5200C.
Conclusions
The results from this work have shown that a loaded vertical column may be unable to bear the applied load at the times predicted using the test results of only unloaded sections. The temperature of hot spots may not be detected by the thermocouples located at the positions required in the draft European test standard prEN 13381-82, which means that a test on a loaded column would be important.
These tests have only been carried out on rectangular hollow sections but they do suggest that similar results could be obtained from circular hollow sections and I-sections. These findings confirm previous testing experience which also demonstrated that hollow sections can sometimes perform worse than similar I-sections because of their shape. It would therefore seem sensible to ensure that the European test standard includes the requirement to test loaded vertical column sections of the different profiles as appropriate.
The results from this work only relate to tests on one particular intumescent protection system for structural steelwork. However, they clearly show that the results obtained using the draft European standard have over-predicted the loadbearing capacity of the sections, when based on testing of loaded beams and unloaded column sections only. The over-prediction of performance by 15 minutes is significant for a product designed to achieve 60 minutes’ fire resistance. These results suggest that the adoption of prEN 13381-82 without the inclusion of a loaded column test could have serious implications for
the structural stability of buildings in fire.
Norman MacDonald is with BRE Global. The author gratefully acknowledges the support of the BRE Trust in funding this project.
References
1. DD ENV 13381-4:2002. Test methods for determining the contribution to the fire resistance of structural members. Applied protection to steel members. BSI, London, 2002.
2. prEN 13381-8. Test methods for determining the contribution to the fire resistance of structural members. Part 8: Applied reactive protection to steel members. Draft dated March 2007.