Calculation - Advanced


» ALWAYS make complete models, with d-f1-f2-f3-f4 junctions and constructions selected for the worksheet.
» ALWAYS check there is an "OK" at the bottom right corner, which means BASTIAN confirms the model is OK.

Are calculations of sound insulation according to EN 12354 consistent with field measurements?

Answer 1: Yes – on the average, provided the calculation model fits he building constructions in situ. BASTIAN does NOT return "safe" values comp to field measurements, i.e. there are NO hidden margins. Users should apply their own margins, according to local legislations and client's specifications.

Answer 2: Estimates of sound insulation according to the EN 12354 usually agree well on the average when compared to field measurements. After a new set of comparisons in May 2013, the input data of concrete walls were increased by 1 dB 2013-06-01. Such changes are rare, this was after more than 10 years of application of the database.

Answer 3: There are some practical aspects you should consider to avoid unnecessary deviations - in both directions.

  1. Field measurements can sometimes reveal worse results than expected. This may for instance be caused by measurement errors or poor workmanship. Here a few well-known shortcomings: air leakage in unintended gaps and cracks reduce airborne sound insulation; flanking transmission via a continuous outer layer of a facade is easy to forget in the calculation model.
  2. Field measurements can also show better results than expected. This may be because you have not customized the calculation model to the actual building. It's not enough to just enter the apartment separating and flanking structures. You should also specify the type of junction of the adjoining structures (e.g. to include structural losses at the junctions of the back-walls). Go to the Extra, Structural Rev Time and adjust the junction types of all sides of all heavy elements.
  3. Errors in input data lead to errors in the calculated sound insulation. Consult the suppliers of the products to be used, in order to provide relevant input data. In case you find flaws in the databases, contact us. An example: impact sound reduction from the laboratory can sometimes be hard to achieve in buildings, especially after some years aging of resilient materials under carpets or parquet floors. Apply some margin at high frequencies if deteriorated elasticity may be reasonable to occur.
  4. Structural losses must be handled: Input data for concrete elements must be converted since the structural losses in a laboratory (for an element itself) are lower than in the current building, which usually gives an increase of 1-3 dB if all junctions are specified. On the contrary, flanking transmission of adjoining structures to the reception room reduces the insulation. Input data for a single concrete element does not tell much about the sound insulation of a building, it is when you put together all concrete elements ("communicating vessels") that you get an estimate of what the building structure performs. All these aspects are handled by EN 12354.
  5. The Swedish building codes BBR and the standards SS 25267/SS 25268 contain rules that say you should fulfill the sound requirement on average within a dwelling or premises, and that single measurements may not deviate by more than 1 or 2 dB. Remember, it is the rejected measurements, or the risk of getting rejected, which determines the minimal construction being appropriate. As a result, the majority of measurements are likely to meet or even exceed the requirement. This may appear as an over-estimation, but is in many cases necessary.
  6. Typically I find 2-3 dB higher sound insulation measured between small rooms than large rooms. Some new test reports show that consultants sometimes only measure between small rooms or only in a favorable direction. The requirement of the standard is clear, test rooms shall be selected to match the building as a whole. In safe designs, sound requirements should be exceeded by 3 dB margin, to obtain reasonable assurance (approximately 90%) to meet the requirements for its field measurement. One or two measurements may still be below the requirement.
  7. If no margin is applied, most of the measurements will still fulfill the requirements but you will get more negative deviations. Only when a large number of measurements is analyzed, these statistical relationships become clear.
  8. Large or systematic discrepancies are probably due to errors in the input data, poor workmanship in situ or in the quality of field measurement. Two common reasons for measurement errors are background noise and insufficient spatial averaging (microphone positions did not cover the entire space).
    Christian Simmons 2013-06-02

What can I do in case I miss a construction in the database?

1. Enter your own data into the database. See the manual, page 169. Basically, go to the Database menue, choose relevant product category and then use right hand click to open a submenue. Either choose NewAsCopy, or NewConstruction. Make sure to use input data for the construction as measured in a laboratory, i.e. without flanking transmission.

2. Contact us. We frequently assess sound insulation of new products for manufacturers and we can certainly help you as well (commissioned work). You can try using Insul, copy and paste, but you should now the limitations of Insul to advise clients on critical constructions using this method (ABK 09 requirement of workmanship). In case you try and use field data, make sure to remove flanking energy before inserting the insulation values, since BASTIAN adds flanking to the model.

3. See below, there are more detailed Q:s on this topic.

I tried to change the C-terms through the Worksheet/Configuration menu, it did not work out. Why?

1. Look at Worksheet/Resulting Diagram/ Xw-button. You then view all combinations of R+Cxx, or Ln+Ciyy.
2. Change preferences through the Extra/Preferences/Performance parameters menu. Worksheet/Configuration is intended to display valid information only. Apo's for this inconvenience.

Is it possible to have different constructions in the source and receiving rooms, e.g. different floors, floorings, ceilings etcetera?

Yes. First, make sure the junctions are correct for the elements you want to apply. Then, d-click in the right column and de-select the Adopt SR-element button to the top-left of the table in the Construction Chooser. You are then free to select any element that matches the selected type of junction, e.g. where there is a mix of heavy and light weight elements.

The results table shows 50.8 dB for L'nw, but the Xw box under the Results diagram shows 50 dB, which seems to be an unsafe way of rounding. Why so?

Changed 2009-02-06: In the Xw box, acccessed from the Resulting Diagram window, all combinations of Xw and C-terms are displayed for convenience, as rounded integers with 1 dB shifts of the reference curve according to the ISO 717 standards. In the table however, decimal Rw and Lnw are calculated to allow for small changes to be studied. Of course, this may cause the wrong impression. The tabulated rounded Xw-values are the correct. Always remember that the Worksheet table values should be rounded downwards (conservatively).

Measured airborne sound insulation between very large rooms is sometimes less than calculated, but the impact sound agrees. Why so?

There are at least three reasons for this: 1) junctions/flanking and structural losses must be modelled correctly, see next question. 2) The data for concrete elements were calculated for ordinary sized rooms. In large rooms (>200 m3), modal coupling may decrease R at low frequency somewhat. Look at the German data for concrete, as these indicate what may happen although these data may appear somewhat exaggerated for many practical situations. 3) Air leakage reduce R at medium and high frequency, this must be prevented at ALL junctions. Prescribe PU or Silicone sealing compound to be filled between all dry elements. Concrete fillers must be specially adapted to make up a tight air sealant.

Light-weight concrete walls, are there any particular pre-cautions to take ?

YES - BEWARE !!! Flanking transmission through thin light-weight monolithic structures may cause severe deterioration of sound insulation if they are connected firmly to the slabs in both ends. Walls, also inner walls, made by l-w concrete, aerated concrete and similar materials with fc in the mid frequency region must NOT be attached firmly to the upper slab. See ph.d. thesis by Kihlman (Chalmers 1960), available at This may now be illustrated in BASTIAN with new element data (no 202 and 203 b-e). In the preference settings, Option tab, activate limits for alpha by the default button (0,050,5). In practice, always use a soft joint to the upper slab, e.g. with mineral wool and a plastic seal to prevent problems. Also avoid firm connections between building service equipment and such walls. To hide pipes in these walls used to be considered a very practical solution, BEWARE... The flanking problem was explained by Tor Kihlman, where the poor R around fc caused both excitation of the wall in the S-room, as well as radiation in the R-room. The bending waves in the wall may cause in-plane waves in the slab concrete floor, which would reduces the junction attenuation compared to the EN 12354-1 value (~high mass ratios). Another transmission may be caused by longitudinal wave resonances in the l-w wall, causing excessive transmission through several joints/floors. The soft joint blocks all of these transmissions, also from various impact sounds (doors, kitchen lockers, WC), but the l-w walls must be completely separated by the soft joints. Take care to insulate any stability reinforcements with rubber pads and prevent any pipes to run through the joints.

How can I model junctions between plasterboard walls, according to the manufacturers?

The predefined junctions in BASTIAN refer to solid junctions between solid elements, which is hardly applicable to common junctions between these double walls, at least not in higher sound classes. Until EN 12354-1 has been updated (work in progress, gently, contribute), the user has to choose a junction without attenuation (#15), and then add their own flanking transmission element where the correct junction attenuation is included. Please contribute examples to the SAU database. However, frequently it turns out this path is not so important, using one of the standard junctions is on the safe side with one exception, timber joist floors on light weight walls, where the ceiling suspension prevents direct transmission but vibration of joists may still excite the wall studs. There are many papers on this issue, look for the AkuLite project reports and congress papers presented at Euronoise or Internoise. Contact me in case of doubt.

The calculated sound insulation of concrete structures sometimes appears to be too HIGH. Why ?

The structural loss factor may in some cases be OVERESTIMATED by the default setting "Maximum coupling". A recent example: A corner room with light weight facade elements on three sides. The default setting will assume a continous slab, although the border of the slab may be considered reflective. Change to User defined SRT and define the types of junction of all connecting elements (4 borders face each element). In practice, this means that second order transmission paths are taken into account with respect to structural loss factor, but they are not used to estimate second order flanking transmission (as follows from the theoretical model in EN12354). Use the Worksheet menu, Export option to view all details in Excel.

Typically, the structural rev. time correction (the "diagram button") is in the range (-2, -4 dB) when several of the borders of the slab do not transmit energy to the surrounding structures. Tell us if you have another experience on this matter. See question "area factor".

How can I model the S & R rooms, when the new rules for V/S=3,1 (m) and V=31 (m3) are applicable according to SS 25267(v3 2004)?

Change the parameters from R'w/L'nw to DnT,w/L'nT,w in the Extra menu, Preferences choice, Performance tab. Bastian now restricts the influence of the V/S and V according to SS 25267, but the room and element border absorption factors are still calculated from the actual dimensions. How the R-to-DnT awa Ln-LnT relate is explained in EN 12354, section 7. NOTE: a common misunderstanding is that DnTw depends on flanking conditions and R'w does not. This is not the case - DnTw and R'w are 100% correlated by the factor V/S, nothing else.

How can I model a) staggered rooms or b) a partition, where only a minor part is common to both rooms (e.g. when a bathroom module blocks part of the common partition) using junctions between light weight constructions ?

EN 12354 does not include staggered junctions between l-w constructions because the transmission loss is not defined. Alt. a: Make a room model that encloses the relevant transmission paths. Turn off walls that do not contribute to the sound transmission (d-click the X-mark to the left in the table). Alt b: Make the model and then assign the actual areas of the direct path and the flanking paths using the geometry control (menu Extra/Geometry). Click and write the actual values in the geometry table. You may make several models to get an idea of the sensitivity of the model. Note: If SS 25267(2004) is applicable, see previous Q. Also note the guidelines about determination of area and volumein the new ISO 140-14.

Row houses, detached by a common thin (10 cm) concrete slab casted on EPS or min-wool heat insulation: The calculated sound insulation appears to be too low, compared to experience. Why ?

A) Up to version 2.2, there was a bug in B. that returns to high L'nw. Upgrade to v2.3.98!

B) Measurements suggest that the loss factor of the plate is much higher than expected just from border absorption of the plates. We suggest that you either a) quick-fix, just insert a +20% thicker slab or b) force one border absorption factor to 0,5 to coop with this. In v 2.3 choose the Extra menu, Struct. Rev. Menu, change fc of each of the borders to 500 Hz. You may also change border absorption in Extras menu. Typically, a 3-4 dB higher insulation should be the result. If more knowledge is gathered on this point, tell us.

Which type of junction is appropriate to describe precast heavy slabs (eg HD/F) on massive wall elements ?

A. Use junction type 9 or 10 to calculate a case with a flexible connection at the junction. Then change to type 1 or 2 to calculate a case with a rigid connection. The most plausible estimate is somewhere inbetween these two cases, as supported by some measurement results. One reason for this is suggested: the precast slabs (e.g. hollow core concrete elements) do not transmit moment (tension fores) since there is no reinforcement in the upper part. There will always be small cracks in the element or the in situ casting. However, shear forces do transmit vibrational power to the other side, causing about 3 dB less flanking transmission than estimated with the junction type 1 or 2, but certainly more than with junction type 9 or 10. There has been one case, where the junction between the slabs (and the holes to some extent) was filled with concrete. The idea was to reduce flanking, but this should rather increase flanking since the mass ratio is then less favourable and the moment is transferred. The measurements indicated indeed, after comparison with another partition cast in the trad. way, that sound insulation was reduced a few dB. In case more knowledge is gathered on this matter, please tell us.
B. HD elements parallel to a massive wall may cause flanking transmission and air leakage. Make sure the junction is properly reinforced and cast with the wall!
C. Details about HDF slabs and walls, look at the information by SvenskBetong

How do I account for the direction of holes in precast heavy hollow slabs in a junction with massive wall elements (perpendicular/parallell to the junction)?

The holes themselves do not influence the sound insulation but of course they ease the weight of the slab. The important thing is the joint. If the elements are parallell to a wall AND the joints between the elements are made in the ordinary way (i.e. they do not transfer moments at high frequencies), flanking sound transmission may be decreased, i.e. the real sound insulation will be higher than calculated. This effect is not taken into account in the database (i.e., a safe approximation). Select junction type according to the above advise. See next question as well. In case more knowledge is gathered on this matter, please tell us.

How do I account for the influence of holes and joints in precast heavy hollow core slabs without topping or with thin toppings without reinforcement? What about air leakage?

The data in the database is already corrected for the influence of joints, cast in an ordinary way. Do not apply forced or restriced border absorption according to the question below, just apply the normal BASTIAN calculation procedure. The results are still within the 3 dB uncertainty interval, this has been tested on a limited number of field measurements. For example, HDF20 330 kg/m2 would be on the limit to pass R'w+C50-3150 52+/-1 dB (class C in SS 25267). Thanks to Spenncon, Aprobo, Strängbetong, Skanska Norge, Bo Gärdhagen and others for their kind support of this investigation. If the HD-elements are parallell to the wall AND the joints between elements are open, i.e. they do not transfer moment, flanking sound insulation may be reduced. This effect is not taken into account in the database (safe approx.). Select junction types as usual. In case the elements are not bonded to each other at the element joints and these are oriented in parallell to the junction, the sound insulation may be greater than calculated. A reason for this is suggested: the hollow core concrete elements do not transmit moment (tension fores) across the joints since there is no reinforcement in the upper part. There will always be small cracks in the element or the topping. However, shear forces do transmit vibrational power to the neighboroughing elements, causing about 3 dB less transmission across joints than estimated. In principle, this applies also to massive precast elements that are jointed without reinforcement. In case more knowledge is gathered on this matter, please tell us.

Another problem related to precast slabs, the risk of air leakage in between the elements (not enough grout filling out the gaps). It may not be sufficient to close the V-grout above the wall, all of the joint must be sealed to prevent air leakage.

How can I calculate heavy double walls as separating walls?

Heavy double walls are only handled in version 2.1 or higher. There is no support on this type of junction in EN 12354-1, annex E. A model has been added to include effects of two sheet, see paper by Metzen and Pedersen at Forum Acusticum 2002. At present time there is little practical experience with the algorithm, but it is well known that the apparent sound reduction of the double wall does not yield in situ because of flanking transmission by the supporting slabs (common to both wall leafs). Double wall elements that rest on resilient material around the perimeter do not interact with the slab and they may be handled as monotlithic walls. Junctions 5 and 6 may be then be applicable, or even 15-16, this depends on the momentum of intertia of the junction between the wall and the slab. These solutions require a high sound insulation of the slabs since there is virtually no junction attenuation. Also see question "area factor".

Can I use only light weight walls and floors? How are laboratory transmission values translated to in-situ values?

Yes, BUT with great care. Radiation from light weight load bearing walls is not included in the model. CEN/TC 126/WG 3 works on this item now to find Kij:s between light-weight junctions. Look at the Vinnova report (SP 2008:16, order at . At InterNoise 2004 (Prague) and Forum Acusticum 2005 (Budapest), some results were presented by Warnock, Quirt and Nightinggale from NRC Canada. Get a new report from (Flanking transmission at the wall/floor junction in multifamily dwellings - quantification and methods of suppression. Research Report 168).

This is a complicated matter. At present time there is little practical experience with the algorithm, but the values obtained in BASTIAN with separated sheets in the junctions seem realistic, even slightly conservative. See note 1 below.

NOTE 1: Floors resting directly on the studs of light weight walls in the recieving room may cause 5-7 dB mor transmission (i.e. less sound insulation) than the laboratory result of the floor alone, i.e. BASTIAN would return 2-4 dB better insulation than achieved. The advise in SS 25267 annex F is to keep at least 4 dB margin between calculation results (incl. flanking paths) and requirements. Additional structual bridges may increase this difference. If this is expected to resemble the field situation, adjust the floor element data accordingly. Open the database, high-light the floor, righ-click and choose NewAsCopy. Change the title and input data. Always contact the manufacturer of the floor system to get additional information on the influence of flanking transmission in typical building constructions, as determined by measurements. The SBUF data (project 11254) on light-weight floors were taken as lab-data, then reduced by 3 dB for these reasons.Contact me in case you need assistance with light weight constructions.

NOTE 2: Correction for reduced flanking below the coincidence frequency fc: Check the sigma_free/sigma_forced correction box (ON) for light-weight constructions, in the Preferences section. Why this? The input data of the separating element typically refer to a case where forced transmission dominates and flanking transmission does not contribute significantly. But when the light elements are used in flanking constructions, the direct path is removed and only resonant transmission occurs. Look at Euronoise 2012; papers by Stefan Schoenwald, Jeffrey Mahn and Cathrine Giogou-Carter. My experience tells that BASTIAN through Heinrich Metzens correction works satisfactory in most cases, compared to Schoenwalds measurements of radiation factors below fc coincidence frequency (cirka –10 dB).

Is the "area factor" according to EN 12354 taken into account, i.e. does BASTIAN correct properly for the increased loss factor due to flanking transmission through connected large heavy structures when only light-weight walls define the rooms ?

Yes & No. BASTIAN does increase the loss factor, but unfortunately not enough if the slab is continous and very large (homogenous concrete) (v 2.3). In case of large heavy walls or slabs with only light weight walls that encloses a much smaller area, e.g. in an office or in a hospital, the vibrational power flow may direct away from the recieving room. The sound insulation horisontally will be underestimated (-1.5,-3 dB) by BASTIAN v 2.3 in extreme cases. For hollow core concrete slabs with a thin topping, the following advice is not applicable unless the joints are bonded in a special way or there is a thick reinforced topping, see question above.

We suggest the use of a practical compromise, as supported by measurement results from the field (*). The loss factor of continous concrete slabs or walls may be increased by the user to get more accurate results. More measurement results would be needed to confirm this way of handling the "area effect" (please contribute!). In the Extras/Structure Reverberation time, first choose Max Coupling. Then choose UserDefined Coupling. Then change fc of each of the connected slabs to 500 Hz along one of its borders, preferably the one at the common junction. This will add 1.5-3 dB horisontally and 0.5-1.0 dB vertically. Use this correction with when YOUR experience tells that the losses at the borders are high, e.g. when the heavy partition wall or slab is much larger (as limited by supporting heavy structures) than the sending or recieving room. Typically, the structural rev. time correction (look at the Diagram) will be in the range (-5, -7 dB) when 3 or 4 of the borders of the slab do transmit energy efficiently to the surrounding structures. (*) Comparison with 7 measured cases in Norway and Sweden (*) have been used to define this procedure. Thanks to contributions from Delta Akustik&Vibration, Brekke&Strand, Sinus, Ingemanssons, WSP and others! NOTE: This does not apply to light weight floorings and sub-floors that are continous under the walls. See paper by Quirt (NRC Canada) at InterNoise 2004.

How can I calculate flanking transmission through heavy sandwich walls (as facades) with BASTIAN?

Use a massive element that is similar to the inner sheet of the sandwich wall, possibly with half the mass of the outer wall added, if the following condition is fulfilled: If the heating insulation is elastic and the outer sheet is comparably weak or is interrupted at the partition/junction. If not, you must ask the manufacturer of the element for the flanking transmission data. Use the appropriate junction type to define the coupling to the inner wall or slab. Assure that the coupling between the elements is properly done at the building site if you depend on this to meet a certain requirement - excess flanking transmission may occur if the blocking effect of the partition fails.

The sending room and the recieving room are shifted or of different size, all heavy constructions. How do I model this ?

We suggest that you make several models, that resembles as close as possible the assumed power flow between the rooms. You may use the staggered junctions, but the experience with these are much less than the simple "cross-shaped" junctions. Apply both, and judge from experience what looks reasonable. It may be necessary to apply a more advanced kind of theory. It may be worthwile to contact us to get a second opinion about the results.

Can I include a room inbetween the sending and the receiving room?

No. EN 12354 include only 1st order transmission paths. Do not add sound insulation values obtained by consecutive calculations - this will certainly overestimate the sound insulation in situ. You may make a rough estimate (probably safe) by choosing a high insulation element or adding an extremely efficient lining, to simulate the blocking effect of the room to the direct sound transmission path but still maintaining the flanking paths.

What if there is a height difference between the sending and the receiving room, e.g. garage-next-to-apartment?

EN 12354 includes only 1st order transmission paths, which means it does not handle staggered junctions (level differences). Using solid junctions gives results on the safe side, say with 3-5 dB on the impact sound transmission. If possible, try and allocate a dilatation joint between the lower floor and the wall to avoid flanking from garage floor to a neighborouging apartment floor.

How can a select an insulated lightweight roof as the exterior element for outdoor sound transmission?

Choose the sloped roof type of calculation sheet. Change the number of exterior elements at the worksheet menu. Change the junction type at all four junctions. Then select the roof and the flanking elements via the CONSTRUCTION CHOOSER dialog box.

Can I change the areas of windows and doors in BASTIAN?

Yes. First of all, you may enter several windows of the same type by right-hand click and Insert a new element row. You may repeat this procedure to add more elements. Or, you may copy and modify the data of a window in the database. N.B. Changing the area from the actual area used in the tests may add an error, typically in the order of 3 dB according to EN ISO 140-3. Larger panes tend to decrease the sound insulation. Finding data for a window with a changed area as compared to the actual test calls for a theoretical approach. We can assist you finding such data based on calculations and measurement results. See also EN 14351.

I do not find my type of external wall in the database - what can I do ?

There exist an infinity of external walls and the database cannot comprise all variations. But the simple answer is to choose one of the variety of walls included in the database with a similar build-up, i.e. number of layers on the inside, same type of studs etcetera. Complex skins on the external side often influence the high frequency insulation but not as much the lower parts that are dominated by mass transmission, i.e. they determine mainly the RA,tr. The more complicated solution is to model a new wall in Insul, copy-and-paste to BASTIAN and reduce 3 dB by the +/- 1 dB button next to the spectral figures. Contact me in case you need assistance (commissioned work).

What safety margin is needed, i.e. what is the confidence in a calculated value using BASTIAN?

There is certainly no general answer to this question that covers all aspects. But before going into long discussions, we recommend all users to keep a margin of at least 3 dB to a required value in an individual case. If some averaging of several cases is allowed, or there is some practical experience with the calculation model for the element used, it may be appropriate to reduce the margin to (say) 1-2 dB. When the main transmission occurs through light weight constructions, such as a timber joist floor, 4 dB seem more appropriate. Some reasons for discrepancies between calculated and measured results are discussed in the body text of the standard (EN 12354). In the NORDTEST tech report 603 (NORDTEST reports.), a comparison between about 40 field results and calculations support this recommendations. Also look at dissertation by Simmons at The data for light-weight floors labelled SBUF xxx have already been reduced by 3 dB, so the 3 dB margin on the overall result should be on the safe side (no guarantees though!). See our paper for a more details.