Floor screeds and finishes

Assessing the performance of a screed that may be showing signs of defect is to some extent a subjective task based on experience and data gathered from site, as well as consideration of the likely uses of the finished floor. However, one of the difficulties faced by the surveyor is the possible variability of mixes and workmanship issues. The lack of information on the quality of preparation, curing and drying, as well as knowledge as to the thickness of the screed, which could vary considerably, also makes assessment problematic.

The first component of a screed investigation therefore involves data gathering in the form of area, use, specification of mix, curing records, and an examination of the construction programme (if available) as well as a preliminary site survey. A sketch plan or general arrangement plan provides a useful base on which to mark crack patterns and details (width, length and direction) as well as areas of hollowness (in the case of a bonded screed), which can be determined by tapping with a suitable implement.

For a new screed, tests to determine whether or not it has lost adhesion should be carried out at least 4 weeks after laying and preferably as late as practicable in the programme to ensure that most drying shrinkage has taken place. Loss of adhesion is not necessarily a problem unless it is over a wide area and is accompanied by unacceptable lifting at joints.

BS 8204-1:2003 provides acceptance limits as to line and level according to future use. This standard may be inappropriate for older buildings constructed prior to 2003 but it nevertheless provides a good guide. Note that flatness and levelness are not the same things - a floor can be flat (i.e. a constant plane) but the plane may not be perfectly horizontal - it could be inclined. The level of the floor can be checked with levels - preferably to millimetre accuracy, but a 2m straight edge should also be used to determine tolerances over a smaller area.

If instructing a level survey, be aware that a close grid yields a very large amount of information often in a form that is somewhat indigestible. A set of spot levels can be difficult to read and involves careful analysis to establish a pattern. Specialist floor surveys are possible to submillimetre accuracy, with survey data produced on a coloured contour plan to make visualisation much more rapid and effective.

Departure from stated tolerances does not necessarily mean that the floor screed should be condemned - this is a judgment call. If the departures are frequent and significant the decision to call for replacement or remedial work is easier.

Aside from dimensional issues, the performance of the screed in terms of crushing strength and propensity to curl or crack must be assessed. Commonly, cores can be taken to examine the degree of compaction, layering, existence of cement balls, mix proportions, aggregate size, etc. Taking cores is preferable to breaking out with a hammer and bolster as it involves less damage to surrounding surfaces. The alternative is to grind out a rectangular test piece, although this is a much more dusty operation (and beware smoke detection systems).

The difficulty with the core approach is that each sample is very local - a large number of test sites would be needed to ensure that representative results had been obtained. Cores can be useful in assisting the identification of failed screeds, BS 4551-2:2005, Mortar. Methods of test for mortar. Chemical analysis and physical testing, provides a methodology for the determination of constituent mix proportions. Under that code, the minimum mass of collected sample per 50m2 of floor area is 100g, made up if necessary by samples of not less than 50g each.

The alternative is to undertake in-situ compression tests in the manner set out in BS 8204-1:2003, Screeds, bases and in-situ floorings. Concrete bases and cement sand levelling screeds to receive floorings. Code of practice. Essentially, the method is based upon the use of a device known as the BRE Screed Tester and the methods first described by BRE in their Information Paper 11/84, BRE screed tester: classification of screeds, sampling and acceptance limits.

The screed tester comprises an annular weight of 4kg or 2kg (depending on the type of screed), which is free to slide up and down a 1m long steel rod. The base of the rod has a surface area of 500mm2 and is placed in contact with the screed. The weight (which is held by a catch) is then dropped down the rod to impose a load on the screed. This test is repeated a further 3 times in the same location. The depression formed in the screed as a result of these repeated shocks can then be measured and compared with published tables to give an indication of the soundness of the screed. While the apparatus does not directly give any information as to mix proportions or any of the other factors described above, experience has shown that as all of these factors are related to performance, measurements to determine the soundness of the screed are highly relevant.

While random readings may give a quick indication of the performance of the screed, it makes sense to establish a regular grid pattern of readings to achieve a more scientific and meaningful result. The grids, or corridors to use the BS parlance, should be at 3-5m intervals (i.e. to give an area of 9-25m2). BS 8204-1:2003, Screeds, bases and in-situ floorings. Concrete bases and cement sand levelling screeds to receive floorings. Code of practice, adopts a methodology that is slightly different to that of IP 11/84 and contains procedures for applying the test to floating screeds, which the earlier guidance did not. Under the standard, random tests are to be carried out in bays (defined by site circumstances or by the division of the floor into a grid). Not less than 3 sets of readings are taken of areas not exceeding 25m2. If faults are found within a particular grid area, the frequency of tests can be increased as appropriate. It is preferable not to test areas within 300mm of any cracks or joints as cracking can result. However, it is important to test areas that will be subjected to high levels of traffic such as doorways and corridors, or where point loads are expected.

BS 8204 provides tables of acceptance criteria - for screeds that are subjected to heavy loads or where surface breakdown is unacceptable, the limit for the depth of the depression caused by the test is 3mm, whereas for lighter duties 5mm may be acceptable. Sometimes, particularly if the floor is worn and/or the anvil strikes a particle of exposed aggregate, the depression depths can be more than would otherwise be the case. A single failure should not be regarded as evidence that the floor has failed the test - the standard allows for an additional 1mm of depression to no more than 5% of the total sample.

The tables provide a good measure for comparison purposes, but it is not simply a matter of selecting an appropriate use category and comparing performance against this. For example, if the screed is to be covered with thick materials such as 25mm woodblocks, natural stone or terrazzo tiles up to 28mm thick, it will be less likely to suffer from damage in use and so some loss of strength and integrity in the screed is less likely to be of significance than where, say, the floor is to be covered with sheet vinyl. Some measure of subjectivity is therefore required, and all relevant factors must be considered before forming a final conclusion.

Floating floors need to be considered as a special case because of the risk of punching shear during the test. If the screed is less than 65mm thick, the results of a compression test may be less reliable - the standard method of dropping a 4kg weight may create a cone shape fracture line beneath the anvil, accompanied by a significant change in tone and additional and sudden indentation. For this reason, particularly with screeds where light uses are intended, a 2kg weight should be used.

The table below gives the maximum permissible indentation limits for in-situ crushing resistance tests, derived from BS 8204-1:2003.

Serviceability is not just confined to measures of crushing resistance; the severity of curling and cracking should also be assessed. Since curling is almost inevitable with some types of screed, a pragmatic approach is needed as to whether or not the movement is acceptable. Just because curling has occurred it does not follow that total replacement is required, unless there is a risk of fracturing under super-imposed loading or where lipping has occurred to an unacceptable degree. Given that all floating or unbonded screeds sound hollow when tapped, undertaking a survey with a hammer or rod may give misleading results. A 2 metre straight edge across joints or cracks may give a better indication of curling in these circumstances.

Aside from cracking arising from physical influences (movements in the structure or substrate) the propensity of a screed to cracking is generally influenced by restraint to early thermal contraction and drying shrinkage. Because it is isolated by a slip membrane, a floating or unbonded screed may be less prone to this type of problem.

Other factors that influence cracking are the water content of the mix. Too much water leads to greater drying shrinkage while certain aggregate types have a high propensity for shrinkage.

Cracks in a screed do not necessarily mean failure, and again need to be interpreted carefully to determine whether they are the result of excess shrinkages or other faults. Fine cracks should not affect serviceability and rarely need filling. Wider cracks need to be assessed in terms of likelihood of continued movement or other effects such as structural movement.

Maximum permissible indentation limits for in situ crushing resistance test