Recommendation for a Standard Rolling Noise Machine
Matthew Edwards, Raimundo Gonzalez Diaz, Nadia Dallaji, Luc Jaouen
RRecommendation for a Standard Rolling Noise Machine
M. Edwards ∗1 , R. Gonzalez Diaz , N. Dallaji , and L. Jaouen Matelys Research Lab, 7 Rue des Maraˆıchers, Bˆat B, 69120 Vaulx-en-Velin, France Aalto University, Department of Computer Science, P.O. Box 13000, 00076 Aalto,Finland Moelven T¨oreboda AB, Bruksgatan 8, 545 31 T¨oreboda, SwedenJuly 2020
Abstract
In the world of building acoustics, a standard tapping machine has long existed for thepurpose of replicating and regulating impact noise. However there still exist other kinds ofstructure-borne noise which could benefit from being considered when designing a building.One of these types of sources is rolling noise. This report details a proposal for defining astandard rolling noise machine. Just as the standard tapping machine can be used in anybuilding and on any surface as a way of characterizing and comparing the performance ofvarious floors with respect to impact noise, the development of a standard rolling devicewould enable the same evaluation and comparison to be made with respect to rolling noise.The hope is that such a prototype may serve as a launch pad for further development,spurring future discussion and criticism on the topic by others who may wish to aid in thepursuit of a truly standardized rolling noise machine.
In the world of building acoustics, a standard tapping machine has long existed for the purposeof replicating and regulating impact noise. This device was originally designed to mimic thesound of human footfall [1], and while this is indeed the strongest source of annoyance amongsurvey responders (in regards to strictly floors) [2], there still exist other kinds of structure-borne noise (whose primary transfer path is through the floor) which could benefit from beingconsidered when designing a building. One of these types of sources is rolling noise.When it comes to assessing the performance of flooring construction in multi-story buildings,the tapping machine is by far and away the most widely used device. This use is justified, tobe sure, as impact noise is often given as one of the greatest sources of annoyance to multi-story building inhabitants [3]. Other sources of noise exist, however, such as rolling noise. Thisis often talked about in regards to outdoor sources such as trains and automobiles, but thereare a plethora of indoor rolling sources which can cause annoyance for building inhabitants aswell: delivery trolleys in commercial spaces, rolling desk chairs in offices (both personal andcommercial), children’s toys, and suitcases, for example. As shown in Figure 1, these itemsgenerate noise due to the small-scale roughness between the floor and wheels, which causesstructure-borne noise to propagate through the building as they roll across the floor.Figure 2 shows the spectra of a typical tapping machine and rolling noise on a classicalconcrete floor, as well as the attenuation of a classical floating floor. The sound signature ofimpact noise is quite different than that of rolling noise in both the temporal and spectral ∗ [email protected] a r X i v : . [ phy s i c s . e d - ph ] A ug igure 1: A typical rolling noise schematic. The small-scale roughness between the floor andwheel generates structural vibrations as the wheel rolls across the floor.domains. Considering the prevalence of tapping machines, this mismatch means that focus israrely given to the effects of rolling noise when designing acoustic treatment systems for floors,resulting in a gap in performance. There is no guarantee that the techniques which are developedto reduce impact noise will necessarily be effective at reducing rolling noise. Furthermore,without a repeatable way of replicating and measuring this kind of noise, the processes to goabout finding solutions to it will remain impeded and incongruent.Figure 2: Comparison of the spectra of tapping noise and rolling noise of a classical concretefloor, as well as the attenuation of a classical floating floor (140 mm concrete slab + adecoupling layer + 40 mm screed). [4]This report details a proposal for defining a standard rolling noise machine. Just as the stan-dard tapping machine can be used in any building and on any surface as a way of characterizingand comparing the performance of various floors with respect to impact noise, the developmentof a standard rolling device would enable the same evaluation and comparison to be made withrespect to rolling noise. It should be noted that, should the proposal be carried forth to thedevelopment of a prototype itself, this device should be treated as just that: a prototype. Thedevelopment of a rolling noise machine that is robust enough to satisfy all the necessary require-ments of being deemed “the standard” is no simple task. As such, one should not expect toaccomplish such a lofty goal on the first attempt. Rather, the hope is that such a prototype mayserve as a launch pad for further development, spurring future discussion and criticism on thetopic by others who may wish to aid in the pursuit of a truly standardized rolling noise machine.If such a device is intended to be as to rolling noise as a tapping machine is to impact noise,then analyzing how the tapping machine came to be should serve as a good way to inspire itsdevelopment. 2 The standard tapping machine: A history
The purpose of this section is not to present a comprehensive overview of the history of thetapping machine. In fact, something of the sort has already been accomplished by Theodore J.Schultz in his extensive report “Impact Noise Testing and Rating – 1980” [5]. On the contrary,this section shall discuss the aspects of the development of the tapping machine which have beendeemed relevant to providing insight into how one may go about developing an equivalent devicefor rolling noise. The history of how the tapping machine become the standard for impact noisemay be used as inspiration for how to go about setting a standard for rolling noise.
The modern tapping machine, a schematic of which is shown in Figure 3, is defined by ISO10140-5 [6]. It has remained essentially unchanged since its first recommendation in 1960 [7].It consists of five hammers, controlled by an AC motor on a cam system, which rise and fall insuccession to generate impact events with the floor upon which the machine rests. The hammerseach have a mass of 500 g (± 2.5%), a fall height of 4 cm (± 2.5%), and are dropped once every100 ±5 ms. Each hammer is supposed to strike the floor only once each time it is dropped(though this has been shown to not always be the case in reality [5]. The hammer materialmay be either brass or steel, with a third option to use a rubber hammer head (the propertiesof which are tightly controlled) on fragile floor surfaces. The hammer heads are to be 3 cm indiameter with slightly rounded faces, with a radius of curvature of 50 cm.Figure 3: A typical tapping machine. The five hammers repeatedly drop in succession togenerate impacts with the floor at a rate of 10 Hz.
The story of how the tapping machine as we know it today came to become the standardfor measuring impact noise in buildings is a long and winding one. Starting in 1927 with thefirst recorded study of footfall noise at the National Bureau of Standards in the United States,and reaching international agreement in January 1960 with the first adoption of the standard bythe International Organization for Standardization (ISO), one may hasten to call the process bywhich the design of the tapping machine was settled upon to be 10% scientific and 90% political.Schultz summed up the major issue that was at hand rather nicely when he stated:“It is worthwhile to recall here that, in acoustical testing, we sometimes haveto choose between making worthless measurements or making meaningless measure-ments. For example, in the case of impact noise: If we do our measurement in the3eld, using some kind of real-life excitation such as walking, the resulting impactsound level may be so low compared to the background noise that it cannot be accu-rately measured; therefore, the measurement is worthless. On the other hand, if wehammer harder (as with the hammer machine), so that the impacts can be readilymeasured, then the measurement is meaningless, because it tells us nothing abouthow the floor behaves under actual use.” [5]Many of the decisions that were made in the design of the tapping machine erred towards beingworthwhile (what exactly this means will be discussed later). As a result, the final product isoften regarded by critics as yielding results which are meaningless. Nonetheless, the tappingmachine we have is the tapping machine that we have: it is the device for measuring impactnoise in buildings, and while there exists much contention over its usefulness, there still appearsto be some agreement that it is still better to have a poor standard than no standard at all.
The mid 1900’s saw a plethora of different tapping machines being developed and testedby various teams across Europe and North America. Most of these drew inspiration from oneanother in some form, and as the years progressed, a regression towards the mean developed, asthe various tapping machines became more and more similar, finally culminating in the standardwe know today. Even from the very first impact test, the design has changed remarkably little.The original tapping machine, developed by the United States National Bureau of Standards,consisted of five rods which could be controlled by a DC motor with a cam system to riseand fall at separate times, roughly once every fifth of a second [8]. From there, other teamsdeveloped their own versions of the device; with the next significant event coming in 1938,when Germany standardized their own tapping device (DIN 4110, 1938) [9]. The differencesacross these devices were found in the varying hammer masses, drop heights, hammer headmaterials, and frequency of impacts. However, the general construction and methodology (i.e.a device which automatically raises and lowers hammers for impacting the floor) remainedessentially unchanged from the start. Perhaps this is not terribly surprising, as it is a relativelystraightforward and intuitive way to generate impacts remotely while conducting measurementsin the room below. Nevertheless, it begs the question of whether other designs were not givencredence simply because of the strong mentality of “this is the way we have always done it” thattends to overpower group decision making, especially in a bureaucratic scenario such as that ofwriting an international standard.As a matter of fact, other variations in the tapping machine design were indeed developedand tested. Furthermore, methodologies for reproducing impact noise which have nothing to dowith the tapping machine at all have even been developed, some of which are currently in usetoday in some parts of the world. Perhaps the most well-known example of this is the rubberball test (Also defined in ISO 10140-5 [6]), which is used in some eastern countries like Japan inplace (or in addition to) the tapping machine, due to its ability to better replicate the sound ofbare feet walking across a floor (a more common occurrence in some eastern cultures). Just asthe name implies, this involves dropping a standardized rubber ball of mass 2.5 kg and diameter18 cm from a height of 1 m multiple times, and recording the resulting impact sound in the roombelow. It can also be used in scenarios where low frequency impact performance is of concern,as it generates an impact noise which has more acoustic energy in the low frequencies than thetapping machine, as exemplified in Figure 4 [10].In setting out to investigate the story of how the tapping machine came to be, we beganwith some sort of an expectation that the development process would have been rooted in deepmathematical analysis. We were hoping to find some sort of documentation throughout theresearch process which detailed the complex theoretical methodologies employed by these earlyresearchers to develop the various tapping machine iterations that were designed throughout4 I m p ac t s ound p r e ss u r e l e v e l ( d B ) ISO tapping machineBang machineRubber ballChild jumping off a chair
Figure 4: Impact sound pressure level on a timber floor from various impact devices, includingthe standard tapping machine and rubber ball. Reproduced from [10].history: with vast numerical explanations which depicted how they arrived at their final proto-types. The reality was surprisingly (or perhaps not at all, depending on one’s familiarity with thematter) nothing of the sort. As far as we could find, the early tapping machines were developedalmost completely by trial and error, with researchers simply trying different materials, weights,drop heights, and methodologies, until they found the one that sounded the best when listenedto from the room below. There was a small amount of theoretical analysis involved, to be fair.For example, two parameters, fall energy (e.g. in [11, 12, 13]) and hammer momentum (e.g. in[14, 15]), were discussed in several of these early designs. Fall energy is usually expressed interms of gram-centimeters, and is defined simply by multiplying the hammer mass by the ham-mer fall height. Hammer momentum on the other hand is defined by multiplying the hammermass by the square root of the fall height. These values were discussed, and a bit of argumentwent back and forth over the years over which one was more accurate at describing the loudnessof a given impact noise (with hammer momentum eventually winning out and being acceptedas the more correct term [15])). However, by far and away, these machines appear to have beendesigned experimentally. It is also worth noting that some of the early test devices were sparsein their details, neglecting to specify aspects of the test procedure (e.g. [16]), or even makingno mention of whether or not they were designed to replicate human footfall noise (e.g. [17]).The variations to the tapping machine itself that were tested during this gestation periodmostly came in the form of differing hammer materials, impact frequencies, and fall energies.The most common materials used started out as wood, rubber, and leather, then migratedtowards brass and steel as the years progressed. Anywhere from between one and six hammerswere used, impacting at frequencies ranging from 4 Hz to 10 Hz. Hammer masses tended to bein the range of 100–1 000 g [5]. Hammer drop heights observed the greatest range of variability,mostly due to the experimentation that was done with measurement procedures which involvedhammer machines with variable drop heights (more on that in the following section).
The divide between low and high fall energy tapping machines is linked to the types ofmeasurement procedures which were tried throughout its infancy. The procedure used todayinvolves placing the tapping machine on the floor of the room above and measuring the soundpressure level in the room below. Thus, the performance depends solely on the sound level in thereception room. However, many early researchers experimented with using a transmission loss5tyle approach, where the sound in both the emission and reception rooms was measured (e.g. in[8, 18, 19]), and the difference between the two used to categorize the performance of the floor.Eventually researchers did move towards only measuring the sound level (or loudness level) in thereception room. Comparative tests do exist today: typically for evaluating the performance of afloor covering. Here the sound pressure level in the reception room is measured with and withoutthe floor covering installed on a concrete slab, and the difference between the two calculatedas the sound attenuation. Nevertheless, the measurements typically remain conducted solely inthe reception room.Subjective listening was also used in place of objective sound level or loudness measurementsin many early studies. This was partially due to the fact that accurate objective measurementequipment had not really been invented yet, but also due to the belief that subjective mea-surements would yield results closer to reality. The two most common subjective methods werethe reference sound method and the just audible method. In the reference sound method, thelistener listened to the tapping machine sound in the reception room and compared it to a cali-brated reference sound. This calibrated reference sound could have been anything from a puretone (the frequency content of which was often never stated in these early reports) to an impulsecaused by passing a voltage spike through a loudspeaker, with the voltage itself measured asthe quantifying value. Examples of this can be found in [11] and [17]. The just audible methodinvolved using a tapping machine capable of dropping the hammers from varying heights, andhaving the listener listen to the impact events over several of these height cycles. By countingthe number of impacts heard between periods of silences (i.e. when the impacts occurring werebelow the threshold of audibility), they could identify the lowest audible drop height. Examplesof this can be found in [1, 14, 20, 21]. It is worth noting that in essentially all of these earlysubjective tests, there was still a degree of calibration involved in the listening. Listeners did notjudge the impact sound while standing in the reception room directly, but listened to it insteadthrough an earpiece which was transmitting the sound from the reception room, and which hadbeen calibrated so that the loudness of the sound through the receiver was perceived to be thesame as the loudness of the impacts while standing directly in the reception room [5].In the wake of World War II in the late 1940’s, a group of research teams met from England,France, Denmark, and the Netherlands to compare test procedures for impact noise [22]. Theyused the same kind of tapping machine and the same test procedure in each of their respectivelabs, but with different measurement equipment (i.e. they each used the measurement equipmentavailable at their respective labs, and did not try to expressly ensure they were identical incalibration). The results were surprising, deviating 10–15 dB from one another, even aftercorrecting for the absorptions of the various lab reception rooms. A second round of comparativetesting was then done, this time using a single tapping machine in a single lab, but each of theresearch groups bringing their own measurement equipment to perform the same measurementprocedure with. This time results were much closer to one another, staying within a range of±2.5 dB after absorption correction. Nevertheless, such a range in the very same lab with thevery same tapping machine was still found to be alarming to the researchers themselves, and acause for concern about the validity of the tapping machine as a reliable test apparatus [5].
Despite the concerns raised, representatives from Denmark, France, the Netherlands, Sweden,and the United Kingdom met in 1948 to discuss adopting an international standard tappingmachine. They agreed to use the already existing German standard tapping machine (defined inDIN 4110 1938 [9]), with the only major change being to use brass rather than wood hammers,with an alternative option to use rubber hammers on fragile floors. This decision was madebased on a desire to reduce as much as possible the variability in measurements, as the materialproperties of brass and rubber can be controlled tighter than wood [5]. The recommendation wasbrought to a symposium of 32 countries later that year, with comments on the symposium code6eing submitted in 1949. A final document on the recommendation for an international tappingmachine was approved shortly thereafter at a meeting in Copenhagen [23]. Schultz also points outthe peculiarity that the German standard was agreed upon being recommended internationallydespite the fact that no representative from Germany was present at the 1948 meeting. Thiscould perhaps lead credence to the “this is the way we’ve always done it” theory being a powerfuldriver in the decision-making process that ultimately resulted in an international standard beingadopted in 1960 which was based on a design created in 1938, despite the fact that numerousalternatives were developed, tested, and proposed during the two decades that separated thetwo.As bureaucratic dealings tend to go, the process was still rather slow in reaching the ISO.The recommendation first came before a committee of the ISO in 1955, with the final draft beingvoted upon in 1958. Of the twenty member countries, this final draft was approved by sixteen,receiving three abstentions and one opposition (from Canada). Finally, in January 1960, theISO officially adopted ISO standard R 140 for conducting impact noise testing (among otherthings), cementing the tapping machine’s standardization in international history [5].
Today there are acousticians who strongly criticize the use of the tapping machine as aneffective means of replicating impact noise in buildings. This is not a recent phenomenon, ascriticisms of the tapping machine have existed since the very beginning [24, 25]. Alternatives havebeen proposed [26, 27, 28], but nothing has been adopted. Schultz points out however, that at thetime the German DIN 4110 1938 standard was made, we already knew essentially all of the sameproblems that we know today, yet they still decided to make the tapping machine the way thatthey did. This does not mean that criticisms of the tapping machine are moot; perhaps it insteadserves as evidence to how heavily politics influence the decision-making process. In regards tousing live walkers as a way to more accurately create standard footfall noise, Schultz states:“It is assumed without question in each study that any machine at all would be preferable tousing live people to excite the test floor (Only in Sippell’s paper (1932) [29] is there a suggestionthat live walkers were used, and even then the evidence is far from certain.).” Ultimately, theinternational symposium in 1948 decided to do what they did because they wanted a way to beable to compare lab tests to one another across locations, and the standard they came up with,however lacking, was the best way to do it.
In this section we shall draw on what was learned from the development of the tappingmachine to inspire the conditions for a standard rolling noise machine. At its most basic level,a standard rolling noise machine should perform a simple function: it should roll across thefloor of the emission room while generating a noise that can be measured in the reception roombelow. There are countless ways by which this can be achieved. The goal here is not to come upwith the design for the prototype itself, but rather to propose a methodology for creating one,and present a discussion of what the priorities should be in the engineering process in order toensure that, when the times comes, the device that ends up being developed is worthwhile.
So, with all of that being said regarding the history of the tapping machine, how does anyof it help inform the decision on how to go about designing a standard rolling noise machine?Looking at the history of the development of the tapping machine, some trends start to emerge.First and foremost, there is a clear preference for repeatability over accuracy. If the deviceyields wildly different results when tested repeatedly in the same scenario, then its usefulness7s a standard is hardly justifiable. Thus, a rolling device should follow suit. A rolling devicerolling across a standard concrete floor in one lab should yield similar measured results to thesame device rolling across a similar concrete floor in a different lab.A second preference can be seen for a machine which requires little or no modification fromlocation to location. Today’s tapping machine generates impacts which, for a lightweight floor,have the capacity to be quite a bit louder than real footfall noise. If one wishes to have adevice which is standard across all floor environments, this is an unavoidable reality. However,this has been deemed to be preferable to the alternative: a device which replicates real footfallnoise perfectly, but is therefore so quiet when being tested on high impedance floor structuresthat the results provide no beneficial meaning. The same should be true of a standard rollingmachine. The device should be capable of generating rolling noise levels which are sufficientlyabove the background noise level for a wide range of flooring constructions, from thick concreteto lightweight timber.During the development of the tapping machine, much focus was placed on choosing theright hammer material. This is a consideration that depends not only on sound, as a leathertipped hammer will generate a slightly different impact noise than a brass tipped one, but alsoon repeatability (linking it with the first point made above). Brass, steel, and rubber can becontrolled much easier than wood or leather, making them more appropriate (in the eyes of the1948 committee) as materials to be used in an international standard device. When looking atrolling noise, such a consideration is equally as important, if not more. The wheels of such adevice should be chosen such that they can be used repeatedly, and for a long lifetime, withoutdegrading to the point of significantly changing the noise being produced. This may provedifficult, as the rolling noise depends on the roughness of the wheel, so any change in the wheelsurface will result in a change in sound. Though perhaps the use of a sufficiently hard material,such as brass or steel, may be able to mitigate these problems to an acceptable degree. This alsohas the added benefit of generating a higher overall sound level, reducing the likelihood that theresults will be too close to the background noise level.Perhaps the most important lesson that can be learned from the development of the tappingmachine is that, no matter what the case, sacrifices will always have to be made. In an idealworld, one could create an impact noise device which perfectly replicates footfall noise in everypossible scenario, while also being repeatable, easily measurable, and standardized. We do notlive in an ideal world: concessions will always be necessary. However, a device which is perhapsless accurate than one would like is better than no device at all. After the end of the secondworld war, acousticians in Europe were eager to establish standards for noise in buildings asquickly as possible, as they did not want buildings which were being constructed to be done sohastily, and with no regards to acoustic quality [22, 30, 31]. Furthermore, they recognized that“more reliable test methods would have to be developed to permit widespread testing of newbuildings” [5]. Today, the time-sensitive pressure to develop standards quickly no longer exists,but the need for test methods which can be used easily in widespread cases remains. A devicewhich is so complex or so modular that it is rendered no longer useful for quick validation testsis hardly worth developing, even if it may satisfy the desires of the most purist of acousticians.With rolling noise, just as with impact noise, the difficulty lies in choosing which corners areworth cutting for the sake of simplicity and which ones are worth preserving for the sake ofaccuracy.
The parameters which dictate the impact sound of a tapping machine are the hammer mass,hammer material, hammer drop height, frequency of impacts, and floor construction. Withrolling noise, the governing parameters are the roughness of the floor and wheel, material prop-erties of the floor and wheel, floor construction, speed of the device, and load on the device.In both cases, the parameters related to the floor are the dependent variables of the equation,8s the whole purpose is to assess their acoustical performance. This may cause an issue forrolling noise. Because the rolling noise depends not only on the intrinsic material propertiesand the macro construction of the floor, but also on its surface roughness, two identical floorswith different surface finishes can yield different noise levels for the same source rolling acrossthem. For example, in the case of rolling office chairs, subjective testing has shown there to bea noticeable difference in the sound produced when changing the flooring material upon whichthe chair rolls [31].As an example, suppose we have a basic two-story structure with a simple concrete slabseparating the two floors, such as the one shown in Figure 5. In the west half of the room, theconcrete remains as it was when it was poured (i.e. relatively rough). In the east half of theroom, a polisher was used to polish the surface of the concrete during drying, resulting in amuch smoother surface. The two halves of the floor remain identical in material compositionand in construction, all that has changed is the surface roughness. In this scenario, a trolleyrolling across the floor from east to west will exhibit a change in sound profile as soon as itcrosses from rolling on the smooth concrete to rolling on the rough concrete (or vice versa). Atapping machine, on the other hand, will sound the same on both surfaces. All else being equal,whether this difference in rolling noise could be large enough to be detectable by a listener inthe room below, or to raise concern for the feasibility of a standard rolling device, is yet to beknown. Nevertheless, it is something that will need to be taken into consideration in the designprocess.Figure 5: Example of a simple two-story concrete structure with varying surface roughness onthe top floor.In turning next to the wheel (or rather, wheels) which will be used on this rolling device, theobvious question is that of what material they should be. While it is true that no single wheelmaterial will be able to encompass the entire realm of possibility, the goal here is to replicate the“worst case scenario”, developing a device which generates as loud a rolling noise as possible (sothat levels on massive floors are still sufficiently high). To this end, a metallic wheel would likelybe the best candidate. Alternatively, just as the tapping machine standard has the option for arubber hammer to be used on fragile surfaces, perhaps two types of wheels may be specified fora rolling device: one soft and one hard.It has been observed through rolling noise measurements that the presence of wheel flatscan dominate over all other influencing factors in rolling noise. A wheel with flat spots rollingon a soft floor covering will generally still be louder than a smooth wheel rolling on bare con-9rete. Thus it may be considered beneficial to use wheels with flat spots on the standard rollingmachine. An additional benefit is that the profile of a wheel is easier to control (from a stan-dardization perspective) than a roughness profile.While use of a flat wheel does have the benefit of generating a louder, more consistent sound,the risk is that the presence of flat spots causes the sound to shift away from sounding likerolling noise and towards sounding like impact noise. This is perhaps analogous to the discussionsurrounding the tapping machine on whether or not it is truly representative of footfall noise.Use of flat wheels should remain an option in the case that a sufficiently high signal to noiseratio cannot be achieved with a smooth wheel, or that a roughness profile is considered toodifficult to tightly control. However, its drawbacks should still be kept in mind.The question of what load should be placed on the rolling device may seem to draw parallelsto the question of what the hammer momentum (or fall energy) should be for the tappingmachine. However, as it has been discovered that the load on the trolley has little influence onthe generated sound level, this is actually only a minor design criteria. The sound level has atendency to decrease slightly with increasing load. The trolley need only be heavy enough toensure that no rattling or other secondary noise is emitted. Beyond that, it may be kept as lightas possible to both increase sound level and reduce the difficulty of moving it around.In a similar vein, the rolling device should be constructed in such a way so as to ensure allwheels always remain in constant contact with the ground. This can be achieved with an eithertwo or three-wheeled device. As shown in Figure 6, moving up to four wheels introduces thepossibility that the device may wobble as it rolls, if a perfect plane is not formed between thefloor and the four contact points. In the interest of automation, a three wheeled device maybe considered preferable over one with two wheels, as the latter would need to be supportedby some other means (likely a person) to avoid falling over (technically gyroscopic stabilizationcould be used on a two-wheel device, but that hardly fulfills the requirement that the designbe simple ). A three-wheeled, motorized design would allow the same flexibility offered by thetapping machine, where it could be set, turned on, and recorded without human interaction.Though considerations would still need to be made to ensure the device does not run into a wall.Fortunately, it has been shown through rolling noise tests that trajectory and the presenceof a human operator provide a negligible change in the radiated sound level. Thus the decisionis greatly simplified as to whether to use two or three wheels, and whether to have the trolleyautomated or manually pushed.The final variable to consider is speed. The sound level and frequency content of the rollingnoise will change with the speed of the rolling device; thus this is a parameter which should beintentionally chosen. Following the goal of replicating a worst case scenario, the speed shouldbe reasonably high such that a high radiated sound level is generated, but not so great as tomake the rolling device unstable or difficult to operate reliably. A speed of around 1 m/s, whichis slightly below the preferred human walking speed of 1.4 m/s [32], may be an ideal balancepoint.
While the focus of this paper is on the development of the device itself, consideration willalso need to be made in generating the method used to measure the rolling noise produced bysaid device. One option would be to use the same ISO standard process already establishedfor measuring impact noise. This has the benefit of being about as straightforward as possible.There is no need to generate a brand new methodology, and rolling noise results may more easilybe compared with those of a tapping machine.. However, it is worth pointing out that even theexisting method of measuring root mean square (RMS) sound levels in the reception room doesnot accurately represent the subjective perception of impact noise either [26, 33]. This is notto say that we may take the opinion of, “well their method has flaws, so it’s okay to put thesame flaws in our method too.” On the contrary, this may serve as an opportunity to correct10 a) One wheel: single point of contactpossible (b) Two wheels: single line of contactpossible(c) Three wheels: single plane of contactpossible (d) Four wheels: multiple planes of contactpossible
Figure 6: Different contact scenarios based on the number of wheels. With four wheels andabove, it is possible to have more than one plane of contact, introducing wobble.the mistakes of yesteryear.Measuring RMS sound levels should in theory provide a better representation of rolling noisethan impact noise, as the former is more homogeneous in nature. Considering the movementof the rolling source across the floor above (contrasted with a stationary tapping machine),directivity may play a role in the perception of the sound in the room below. This is dealtwith in the standards surrounding the sound emission of earth moving equipment by havingthe machine drive through a microphone hemisphere at a constant speed, and measuring theaverage sound power level of the hemisphere from the time the center of the machine entersone side of the hemisphere to the time it exits the opposite side [34]. Additionally, for testingin reverberant rooms, a microphone on a rotating boom is often used. While the presence of amulti-floor setup with indoor rolling noise prohibits the use of a microphone hemisphere aroundthe rolling source (not to mention the headache that would cause for acousticians, should sucha method be adopted), the fact remains that averaging of the recorded sound level over theduration of the rolling event as the device moves from one side of the floor to the other mayserve well to remove any discrepancies caused by directivity. Fortunately, this is precisely whatis done in ISO 10140-5: either a number of microphones spaced apart or a single microphone ona rotating boom are used in the reception room. Nevertheless, a perceptive test which comparesrolling noise recordings captured with different methods could yield beneficial information abouthow well these methods correspond to what humans actually perceive..Finally, in order for measurements to remain comparable across laboratory locations, it isadvised that a consistent thickness be used for the base concrete slab upon which the floor rolls.As 140 mm is a typical thickness for concrete floors in multi-story buildings, specifying thatall rolling noise laboratory measurements be conducted using a base floor of 140 mm concretewould allow for easy comparison across a number of laboratories.11
Proposed development process
With all the above being said, the proposed development process for a standard rollingmachine prototype could be considered as follows:1. Conduct a series of tests on various rolling products in a two-story reverberation chamberin order to gain a wide range of audio samples for the different kinds of rolling noise whichmay be found in indoor scenarios, thus complimenting the existing indoor rolling test data.2. Perform a subjective study to identify how varying the measurement method of the abovetests changes the perceived rolling sound compared with what a listener actually hears.3. Use the results of the rolling product tests to aid in the development of specific designrequirements for the prototype. Choose dimensional and material criteria in such a wayas to provide “worst case scenario” representation of the range of sound profiles observed.4. Develop a theoretical prototype design using a computer-aided design (CAD) software.5. Fabricate first prototype. Test the device in a two-story reverberation chamber, comparingthe results to those of the previous tested rolling products. Compare both the soundgenerated by the prototype, as well as the applicability of the measurement method decidedupon in step 2.6. Revise the prototype design and measurement method as necessary based on the outcomesof the test.7. Disseminate results.
This report details a proposal for defining a standard rolling noise machine. Just as the stan-dard tapping machine can be used in any building and on any surface as a way of characterizingand comparing the performance of various floors with respect to impact noise, the developmentof a standard rolling device would enable the same evaluation and comparison to be made withrespect to rolling noise. Throughout the research process, it was discovered that a preferencefor high repeatability over high accuracy was continuously made in the decisions that went intothe standard tapping machine design. These decisions may be used as guidance and inspirationin the future development of a standard rolling machine, with a possible goal being to developa prototype which may be used as a reference device for conducting rolling noise measurementsin buildings.
The work which went into this report was conducted as part of the Acoutect program inconjunction with NCC during the demonstrator group 3 secondment, November 5–16, 2018. Theauthors would like to thank NCC for hosting the team during the duration of the secondment:notably Linda Cusumano, Christina Claeson-Jonsson, and Birgitta Berglund. This project hasreceived funding from the European Union’s Horizon 2020 research and innovation programmeunder grant agreement No 721536.
References [1] H. Reiher, “¨uber den Schallschutz durch Baukonstruktionsteile (Sound Isolation by BuildingConstruction Elements),”
Beiheft zue Gesundh.-Ing. , vol. 2, no. 11, pp. 2–28, 1932.122] K. Jarner¨o, D. Bard, and C. Simmons, “Vibration Performance of Apartment Buildings withWooden Lightweight Frame-Work – Residents Survey and Field Measurements. AkuLiteReport 6,” Tech. Rep. 2013:17, SP - Sveriges Tekniska Forskningsinstitut, SP Tr¨a, 2013.[3] J. Forss´en, W. Kropp, J. Brunskog, S. Ljunggren, D. Bard, G. Sandberg, F. Ljunggren,A. ˚Agren, O. Hallstr¨om, H. Dybro, K. Larsson, K. Tillberg, K. Jarner¨o, L.-G. Sj¨okvist,B. ¨Ostman, K. Hagsberg, ˚A. Bolmsvik, A. OIsson, C.-G. Ekstrand, and M. Johansson,“Acoustics in Wooden Buildings, State of the Art 2008,” Tech. Rep. 2008:16, SP TechnicalResearch Institute of Sweden, 2008.[4] M. Edwards, F. Chevillotte, L. Jaouen, F.-X. B´ecot, and N. Totaro, “Rolling Noise Modelingin Buildings,” in
Internoise 2018 , (Chicago, IL USA), Aug. 2018.[5] T. J. Schultz, “Impact Noise Testing and Rating - 1980,” Final NBS-GCR-80-249, BoltBeranek and Newman Inc., Cambridge MA, USA, Jan. 1981.[6] “ISO 10140-5:2010 - Acoustics – Laboratory Measurement of Sound Insulation of BuildingElements – Part 5: Requirements for test facilities and equipment,” 2010.[7] “Field and Laboratory Measurements of Airborne and Impact Sound Transmission,” 1960.[8] V. L. Chrisler and W. F. Snyder, “Transmission of Sound Through Wall and Floor Struc-tures,” Tech. Rep. 48, J. Research NBS 2, 1929.[9] “DIN 4110, 1938: Technischen Bestimmungen f¨ur die Zulassung neuer Bausweisen (Tech-nical Recommendations for Licencing of Building Constructions),” Tech. Rep. DeutscheIndustrie Normen 4110, Abschnitt D 11 (Schallschutz) (Section D11 - Sound Protection),1938.[10] C. Hopkins,
Sound Insulation . Amsterdam: Elsevier / Butterworth-Heinemann, 1st ed ed.,2007.[11] “National Physcial Laboratory Annual Report,” Tech. Rep. III - Sound A.-General Researchand Maintenance of Standards, (1) Measurement of Sound Transmission Through BuildingStructures (b) Floors, 1935.[12] A. Gastell, “Schalld¨amn-Messungen in der Fraxis und Vorschl¨age zue Normung desSchallschutzes von Wohnungstrennw¨anden und Decken (Sound Isolation Measurements inFractice and Proposals for Standardization of Sound Protection of Walls and Floors inDwellings),”
Akustische Zeitschrift , vol. 1, pp. 24–35, 1936.[13] W. Schule, W. Bausch, and R. Seeger, “W¨armeund Schalltechnische Untersuchungen an derVersuchssiedlung Stuttgart-Weissenhof (Thermal and Acoustical Researches at the Weis-senhof Housing Development, Stuttgart),”
Gesundheits-Ingenieur , vol. 60, pp. 709–713,Nov. 1937.[14] G. Hofbauer, “Der Schallschutz von Decken (The Sound Isolation of Floors),”
Gesundheits-Ingenieur , vol. 54, no. 42, pp. 562–566, 1934.[15] W. Bausch, “Senalld¨ammungsmessunnen im Laboratorium und in fertigen Geb¨auden(Sound Isolation Measurements in the Laboratory and in Finished Buildings),”
Beiheftzue Gesundh.-Ing. , vol. 2, no. 20, 1939.[16] E. Meyer, “Schallisolation und Schallabsorption: Verfahren zu ihrer Messung (Sound Iso-lation and Sound Absorption: Procedures for their Measurement),”
Zeitschr, VDI , vol. 74,pp. 273–279, Mar. 1930. 1317] E. Meyer, “Schallschutz in Hochbauten (Sound Protection in High Buildings),”
Zeitschr,VDI , vol. 57, no. 1, pp. 563–566, 1931.[18] “Measurement of Sound, Noise Abatement, Acoustics of Buildings, Sound TransmissionThrough Walls, Floors, and Pipes,” Tech. Rep. Engineering 148, National Physical Labo-ratory, England, Nov. 1939.[19] R. Lindahl and H. J. Sabine, “Measurement of Impact Sound Transmission ThroughFloors,”
The Journal of the Acoustical Society of America , vol. 11, pp. 401–405, Apr. 1940.[20] F. M. Osswald, “Method for Measuring Sound Isolation, in Particular of Impact Sound,”
The Journal of the Acoustical Society of America , vol. 7, pp. 261–264, Apr. 1936.[21] P. Kipfer, “Insulation Measurement of Footstep Noises,” tech. rep., The Physical Society,London, 1949.[22] C. Kosten and J. van den Eijk, “Comparative Impact Sound Measurements,” tech. rep.,The Physical Society, London, 1949.[23] P. Parkin, “Provisional Code for Field and Laboratory Measurements of Airborne andImpact Sound Insulation,” tech. rep., The Physical Society, London, 1949.[24] T. Mariner, “Comments on Impact-Noise Measurement,”
The Journal of the AcousticalSociety of America , vol. 35, pp. 1453–1453, Sept. 1963.[25] T. Mariner, “Criticism of the ISO Impact-Noise Test,”
The Journal of the Acoustical Societyof America , vol. 36, pp. 1024–1024, May 1964.[26] T. J. Schultz, “Alternative test method for evaluating impact noise,”
The Journal of theAcoustical Society of America , vol. 60, pp. 645–655, Sept. 1976.[27] J. J. LoVerde and D. W. Dong, “A dual-rating method for evaluating impact noise isolationof floor-ceiling assemblies,”
The Journal of the Acoustical Society of America , vol. 141,pp. 428–440, Jan. 2017.[28] W. Scholl, “Impact Sound Insulation: The Standard Tapping Machine Shall Learn toWalk!,”
Building Acoustics , vol. 8, pp. 245–256, Dec. 2001.[29] K. Sippell, “Schallschutz durch Baukonstruktionsteile (Sound Protection by Building Con-struction Elements),”
Gesundheits-Ingenieur , vol. 55, pp. 535–540, 1940.[30] L. Cremer, “Stand der Forschung im Baulichen Schallschutz (Status of Research of NoiseControl in Buildings),”
VDI-Zeitschrift , vol. 91, pp. 199–205, May 1949.[31] D. Dal Pal`u, E. Buiatti, G. E. Puglisi, O. Houix, P. Susini, C. De Giorgi, and A. Astolfi,“The use of Semantic Differential Scales in Listening Tests: A Comparison Between Contextand Laboratory Test Conditions for the Rolling Sounds of Office Chairs,”
Applied Acoustics ,vol. 127, pp. 270–283, Dec. 2017.[32] R. C. Browning, E. A. Baker, J. A. Herron, and R. Kram, “Effects of obesity and sex onthe energetic cost and preferred speed of walking,”