methodology for design of a vibration operated valve for abrasive

8/10/2019 Methodology for Design of a Vibration Operated Valve for Abrasive

1/12

Methodology for design of a vibration operated valve for abrasiveviscous uids

Khashayar Behdinan a ,1 , Behrokh Khoshnevis b ,a Chevron PLCs Plus Int., Chevron North America Exploration & Production, San Joaquin Valley Business Unit Facilities Engineering, Bakerseld, CA 93311, United Statesb Epstein Department of Industrial and System Engineering, University of Southern California, Los Angeles, CA 90089-0193, United States

a r t i c l e i n f o

Article history:Received 1 January 2013Accepted 9 July 2013Available online 4 September 2013

Keywords:ValveVibrationPiezoAbrasive viscous uid

a b s t r a c t

In this paper a new ultrasonic operated valve is presented. The ultrasonic valve design was analyzed andthe valve was prototyped and tested for ow control of abrasive viscous uid. This innovative valve con-cept is based on controlling the friction of material by employing several friction elements along the owdirection. Abrasive particles in the viscous uid are stopped by the force of friction when coming intocontact with the friction elements. Friction is neutralized by use of vibration to break away the abrasiveparticles from the friction element surfaces. Several factors were considered in designing the piezoelec-tric valve. Factor identication was done by conducting experiments and analyzing the resulting data.Some important factors that affect the valve design were recognized to be pumping pressure, size of fric-tion blades along the direction of ow, density of material, viscosity, amplitude of vibration, frequency of vibration, and proportion of abrasive particles in the mix. First, a method was designed for measuring thefriction coefcient of the given viscous materials. A design of experiment approach was pursued in orderto identify the signicant parameters. A piezoelectric transducer was used, which vibrated at the reso-nance frequency of 20 kHz. FEM modeling was used at that stage to ensure that the resonance frequencyof the designed valve matched the resonance frequency of the transducer and booster assembly that pro-vided vibration. In order to perform proportional ow control pulse width modulation was used to con-

trol the duty cycle of theultrasonic power transferred to thevalve. A study was performed to ndthe bestvibration characteristics.

2013 Elsevier Ltd. All rights reserved.

1. Introduction

Controlling the ow of abrasive viscous uids has been a majorproblem, especially when the uid contains relatively large parti-cles of different sizes. Common problems include clogging, jamming of movable components of conventional valves, andcleaning of the valve after use. The challenge of this study, there-fore, was to design a solid state (i.e., solidly connected parts) vibra-tion-operated valve that would provide steady, on-demand ow

for abrasive viscous uids at the desired rates, based on the follow-ing given information:

1. Required ow rate.2. Material friction characteristics.3. Desired pumping pressure.4. Pipe diameter, which affects the design of the starting section of

the valve.5. Material density.

This research has been motivated by the Contour Crafting auto-mated construction project in which accurate ow control of viscous concrete is necessary. Concrete is composedof sand, gravel,cement, water and chemical admixtures. Sand and gravel compo-nents of concrete are often very abrasive. The solution describedhere has worked very well for concrete ow control and we pre-sume it should work for a wide variety of other viscous compositematerials that may contain granular aggregates and/or bers.

1.1. Previous attempts at controlling cementitious uid ow

In the Contour Crafting automated construction project [6] ,where control of concrete ow is a strict requirement, severalattempts have been made to use available solutions or devisenew solutions for ow control. First, a relatively small, progressingcavity pump was selected, which is shown in Fig. 1.1 . When thispump is used the ow pulsates as a result of the sequential outputof discrete volumes trapped in the helical cavities of the rubberstator. Furthermore, the pump easily stalls if the proportion of sandexceeds a certain limit. Also the assembly is very heavy (nearly70 kg) and hence is not suitable to be attached to the nozzle assem-bly which is supposed to be moved swiftly in the 3D space.

0957-4158/$ - see front matter 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.mechatronics.2013.07.003

Corresponding author. Tel.: +1 213 740 5255.E-mail addresses: [email protected] (K. Behdinan), [email protected] (B.

Khoshnevis).1 Tel.: +1 310 402 6304.

Mechatronics 23 (2013) 10251036

Contents lists available at ScienceDirect

Mechatronics

j ou r n a l ho m e p a g e : www.e l s e v i e r. c o m / l oc a t e / m e c ha t r on i c s
http://dx.doi.org/10.1016/j.mechatronics.2013.07.003mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.mechatronics.2013.07.003http://www.sciencedirect.com/science/journal/09574158http://www.elsevier.com/locate/mechatronicshttp://www.elsevier.com/locate/mechatronicshttp://www.sciencedirect.com/science/journal/09574158http://dx.doi.org/10.1016/j.mechatronics.2013.07.003mailto:[email protected]:[email protected]://dx.doi.org/10.1016/j.mechatronics.2013.07.003http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://crossmark.crossref.org/dialog/?doi=10.1016/j.mechatronics.2013.07.003&domain=pdfhttp://-/?-


2/12

A new dosing device was then invented by the authors. This de-vice was signicantly smaller and lighter than the progressing cav-ity pump system. This dosing device had a rotating shaft with aninside cylinder. A rubber acted as a piston inside the hollow shaftand by each rotation the concrete was dosed and delivered. Thisdosing device, shown in Fig. 1.2 , performed its function very satis-factorily but the rubber components used wore out rapidly.

Use of proportional valves with feedback control was then con-sidered. Subsequently, several valves were designed, fabricatedand tested. For example, a pinch valve concept ( Fig. 1.3 ) using a la-tex tube, was devised. The tube was mechanically pinched by ablade that could reciprocate by means of a DC motor connectedto a worm gear. The valve operated somewhat satisfactorily buthad the occasional problem of segregation and clogging.

After testing the pinch valve, a new cut-off valve was designedand built. In this valve, pinching was avoided and instead a bladewas implemented in the valve to obstruct the concrete ow. A pic-ture of this valve is shown in Fig. 1.4 .

This valve had the problems of cleaning and blockageand it per-formed only as an open-close valve, that is, the valve was not ableto regulate the ow and sand particles tend to obstruct the valvesmoving parts.

2. Mobility of fresh concrete

Concrete is a mix composed of a uid phase and a solid phase.The uid phase is the cement slurry and sand is the solid phase.

In mixed concrete particle assembly is composed of non-cohe-sive particles (aggregate grains) and cohesive particles (cement

grains) surrounded by mixing water membranes. In the static stateboth particle groups are only subjected to the frictional resistance,but in the moving state only the former simultaneously bears theviscous resistance together with the frictional resistance.

2.1. The ow of concrete along a pipeline

The ow of concrete along a pipeline may be visualized as var-iously sized aggregate and cement particles suspended in water.Ferraris [8 ,9] showed that freshly mixed cement paste exhibitsnon-Newtonian uid properties, in that it has a yield point and aplastic viscosity which varies both with time under shear and alsoshearing rate. The addition of aggregate to cement paste further re-moves concrete ow properties from the Newtonian concept.When starting with new pipes it must be remembered that theinternal surface of these pipes will be relatively rough comparedwith pipes that have been used for some time.

Loadwick [15] discussed mobility of fresh concrete in terms of its viscosity, cohesion, and internal resistance to shear. Brown(1977) created a model to relate the state of concrete in the pipe-

line to the mixed components. He established a pumping systemtopump the concrete along a pipe and measured the mobility andpumpability of fresh concrete. He observed that a concrete mixturewith excessive coarse aggregate results in a loss of cohesion andmobility. Base on his study, aggregate particle shape and size dis-tribution is important factors inuencing the rheology of the mix.Practical experiments with concrete exiting a pipeline show thatit ows in the form of a toothpaste-like plug separated fromthe pipe wall by a thin lubricating layer ( Fig. 2.1 ).

Practice has shown that pumpability of concretedepends on theproportions and aggregate grain sizes (coarseness or neness) ( Ta-ble 2.1 ) of the mixture. This means that the cement content is of major importance and, in fact, low cement concrete is not consis-tently pumpable. ( Fig. 2.2 ).

Movement of the concrete along the pipeline is resisted by thepipeline itself. This ow resistance consists of hydraulic shearing of the lubricating layer, and solid friction of particles against the pipewall.

Fig. 2.3 shows that ow resistance will be low if concrete is inhydraulic (saturated) state. However, if the permeability of con-crete at the pressure gradients applied by the pump is such thatwater ows at an excessive rate down the pipeline, then the seg-ments of concrete near the pump may become dewatered. Thisdewatering process could transform the concrete from its hydrau-lic state to the transitional or friction state, with a correspondingdramatic increase in ow resistance. The pressure levels requiredto maintain a given ow rate are dependent on the ow resistance.Both permeability and ow resistance are functions of the mix pro-portions. Additionally, permeability is governed by pressure gradi-ent which, for a constant diameter pump, is directly proportionalto the ow resistance.

The ow of concrete in a pipe depends on a variety of factorsincluding viscosity, yield stress, size distribution and the shape of coarse aggregates. Modeling the ow of complex uids like con-

Fig. 1.1. A progressing cavity pump driven by a large DC motor.

Fig. 1.2. A new dosing pump coupled to its driving DC motor.

Fig. 1.3. Structure of a pinch valve and rupture of the latex tube due to valve failure and subsequent pressure accumulation [11] .

1026 K. Behdinan, B. Khoshnevis / Mechatronics 23 (2013) 10251036


3/12

crete presents a great research challenge because of the necessityto account for the motion of the aggregates.

Almost all particles make contact with adjacent particles whilethe uid is in owing state ( Fig. 2.4 ). The particles that are com-pletely or partially separate from other particles in the stagnationstate will push aside their surrounding water and make contactwith other particles once the ow starts ( Fig. 2.5 ).

According to Murayama (1991), the contact slant of particles isnot generally parallel to the maximum shear (MS) plane andchanges with the particle positions. The contact between particlesresults in the inter-particle friction that is thought to followCoulombs solid friction law. Hence, whether the particles in freshconcrete are stationary or moving, all the cement particles and allthe aggregate particles are subject to frictional resistance. Depend-ing on how the structure responds to an applied shear stress, onecan observe different types of macroscopic ow behavior such asyield stress behavior and viscoelasticity.

2.2. Behavior of fresh concrete during vibration

Ferraris [10] performed research into testing and modeling of fresh concrete rheology. Rheology is the science that deals withthe ow of materials and in this case includes the deformation

characteristics of Semi-hardened concrete.At low stresses, the material behaves as a solid of extremelyhigh viscosity. As stresses increase, concrete behavior graduallychanges to that of a liquid ( Fig. 2.6 ).

Ferraris [10] states that ow is restricted by frictional, cohesive,and viscous forces. Cohesion develops due to attractive surfaceforces between particles while resistance is caused by the viscousow of the matrix [1] . No ow occurs when increasing the shearstresses below the yield value and the concrete behaves like asolid. At higher oscillating stresses the bond strength betweenparticles becomes insufcient to prevent ow, and at the sametime the viscosity gradually decreases. Concrete mixture propor-tioning, therefore, indirectly takes into account that the viscosityof the lubricating cement paste can be adjusted to the vibratorystress and its frequency. It follows that with increased vibratoryor consolidation pressure an increase in paste viscosity is required,i.e., a decrease in the watercement ratio and/or an increase infrequency of vibration is required [11] .

2.3. Fundamental mechanism of vibration valves

The novel vibration valve operation principle is based on thefriction between abrasive particles within the uid material andthe inner walls of the ow conduit. Thevalve, therefore, works onlyfor viscous materials that contain abrasive particles. Under thepressure applied to cause ow of the viscous uid, abrasiveparticles accumulate along the ow path due to their inter particlefriction with the conduit inner surfaces and dewatering. Movingthese particles needs vibration to allow the passage of material.Vibration of the conduit walls breaks this mechanical locking.

Fig. 2.7 shows the friction concept between particles and theconduit walls. In this gure the friction caused by interlocking of abrasive particles and conduit surface is magnied by the bridg-ing phenomenon. The pressure of incoming material against themiddle particles of the arc formation causes the particles nearthe conduit surface to be pushed against the surface, henceincreasing friction. Vibration of the conduit walls breaks awaythese mechanical locking, hence allowing the ow.

3. Power ultrasonic

The eld of power ultrasonic, which represents an important

eld of industrial electronics, has experienced very swift anddynamic development in the past two decades. As a result, some

Fig. 1.4. Mechanical parts of the cut-off valve.

Fig. 2.1. Concrete velocity prole along pipe. Source: Ede, A.N., The resistance of concrete pumped through pipelines.

Table 2.1Proportion of each constituent normally used in concrete.

Cement Aggregate Water

Fine Coarse

Volume proportion range 516% 2229%

4254%

1220%

Weight proportion range for unitcement weight

1 0.72.4

1.34.6

0.40.7

Fig. 2.2. Pressure to overcome friction. Source: Ede, A.N., The resistance of concretepumped through pipelines.

K. Behdinan, B. Khoshnevis/ Mechatronics 23 (2013) 10251036 1027
http://-/?-http://-/?-http://-/?-http://-/?-


4/12

systematic design and construction methodologies for new ultra-sonic devices have been developed. Applications in the eld of

power ultrasound have extended into different branches andprocesses of many industries, including mechanical, electrical,and chemical. Alongside new applications of ultrasound ever new-er and more efcient sandwich transducers are being designed anddeveloped, and many scientic papers have been publishedpresenting various aspects of power ultrasonic techniques, espe-cially different electromechanical models that represent optimaldesign of ultrasonic transducers. Following are some examples of the related developments.

3.1. Ultrasonic acoustic vibration generation

The designed valve was connectedto an ultrasonic converter via

an interface. The best operating frequency of the ultrasonicconverter is normally when the maximum traveling-wave

amplitude is reached and when a relatively stable oscillation isestablished.

The best operating ultrasonic systems are those that producevery strong mechanical oscillations or high and stable vibratingmechanical amplitudes, with moderate electric output power fromthe ultrasonic power supply. The second criterion is that thermal

power dissipation of the total mechanical system, in continuousoperation with no additional system loading, be minimal.

Fig. 2.3. Different states of pumping concrete. Source: F. Loadwick, Some factors affecting the ow of concrete through pipeline.

Fig. 2.4. Model of particles in the ow path. Source: http://ciks.cbt.nist.gov/garbocz/scc_2003/node1.htm.

Fig. 2.5. Model particles contact. Source: Murayama, 1991.

Fig. 2.6. Viscous ow frictional resistance. Source: ACI.

Fig. 2.7. Friction force between abrasive particles.



5/12

After selecting the proper sandwich transducer, the boostermust be selected. The booster in this case attaches to the valvehead and delivers power to the load.

3.2. The main components in ultrasonic acoustic vibration generation

The main components in ultrasonic acoustic vibration genera-tion are the following ( Fig. 3.1 ):

(A) Ultrasonic generator, being an electronic ultrasonic signalgenerator. The converter, which is an amplication devicefor generating high power oscillating electric current froman electrical signal, is driven by an electronic signalgenerator.

(B) High power ultrasonic converter, which converts electricalenergy into high frequency mechanical vibration.

(C) Booster, being a metal bar of, for example, aluminum ortitanium, that as a mechanical wave guide connects theultrasonic transducer with an acoustic load, oscillating body,or resonator; this can also boost the amplitude of input

signal.(D) Acoustic load, which is the mechanical resonating body.

The acoustic load, which in our case is the designed valve, wasdriven by incoming frequency and amplitude modulated pulse-train, causing it to begin oscillating in one or more of its naturalvibration modes or harmonics.

3.3. Selecting sandwich transducers

The transducer converts electricity into high frequencymechanical vibration. Transducer active elements are usually pie-zoelectric ceramics although magnetostrictive materials are alsoused. Transducers are also called converters. Bolt-clampedLangevin-type (sandwich) transducers are widely used as efcientvibratory sources in various elds of industrial application of high-power ultrasonics. A transducer of this type can steadily generatehigh-amplitude ultrasonic vibrations ( Fig. 3.2 ).

By applying a sine wave voltage to the piezoelectric ring thethickness of the piezoelectric rings is periodically altered. This

oscillation of the thickness of the piezo rings causes longitudinalvibration of the transducer [27] .

Fig. 3.1. Schematic of generator, converter and load. Source MPI.

Fig. 3.2. A sandwich transducer. Source MPI.

Fig. 3.3. A typical booster. Source Branson.

Fig. 3.4. Different booster types. Source: http://www.staplaultrasonics.com/c3-struc/struc3.htm.



6/12

The amplitude of the transducer ultrasonic oscillations is

between 0 and 20 l m. The largest amplitude develops at the trans-ducer tip at the longitudinal resonance of the horn. The frequencyto which the horn is excited is determined by the shape of the hornand type of materials used to make the horn.

3.4. Selecting the booster

In order to amplify mechanical vibration, a booster is used. Thebooster adjusts the vibration output from the transducer andtransfers the ultrasonic energy to the horn. The booster also gener-ally provides a method for mounting the ultrasonic stack to asupport structure ( Fig. 3.3 ).

Different boosters exist for different applications with varyingproperties that could increase or reduce the vibration amplitude

(Fig. 3.4 ).The graph of a step-up booster is shown in Fig. 3.5 .

In our application a 3 KW transducer was selected which couldgenerate enough vibration to directly operate the valve, hence a1:1 ration booster was selected which merely served as a couplingbetween generator and horn. Also the mounting bracket wasinstalled on the booster so that it could be easily connected to axture without transferring vibration to the machine body.

3.5. Selecting an ultrasonic generator

An ultrasonic generator is a device that generates ultrasonicwaves in form of electric current. The ultrasonic waves could be

Fig. 3.5. Vibration amplication diagram of a step-up booster. Source: http://www.staplaultrasonics.com/c3-struc/struc3.htm.

Fig. 3.6. An ultrasonic generator with transducer and booster placed on top. SourceBranson.

Fig. 3.7. Frequencyresponse of a piezoelectric transducer. Source: Miodrag ProkicMPI Company.



7/12

frequency or amplitude modulated to produce variable outputwaves. A frequency modulated ultrasonic wave could also beamplitude modulated. This is the most general case of designerwaveform for a single generator driving a single transducer array.

Resonance phenomena are caused by repeatedly pumpingenergy into the part to be vibrated at the known resonancefrequency of the part ( Fig. 3.6 ).

In order to operate our choice of single piezoelectric transducerin resonance a factory-set generator with a frequency windowcontaining 20 kHz (e.g., 1822 kHz) was selected for driving thevalve. Having a wider interval of carrier frequency range wouldnot have offered additional benet in this application and wouldpossibly destroy, overload, or over-heat the mechanical systemby trying to drive it in opposition to its acoustical nature.

3.6. Vibration modes

Every elastic mechanical system, body, or resonator that canoscillate has many vibration modes as well as frequency harmonicsand sub harmonics in the low and ultrasonic frequency domains.

Many of these vibration modes can be coupled acoustically andmechanically, while others would stay relatively independent.

Fig. 3.7 shows a typical diagram of a transducers mechanicalvibration mode. In the valve application case it was possible tosimulate many of the mechanical vibration modes of the chosentransducer by means of FEM simulations. FEM modeling enablesthe designers to determine the vibration modes that are presentin a given frequency range, and to accurately characterize thecorresponding displacement elds. In reality, only a few of thesecalculated modes are excited with signicant amplitude.

By measuring the frequency response, we can see that, in addi-tion to the main longitudinal resonance frequencies fs and fp, aphysical transducer possesses many other secondary resonancesthat arise from other mechanical modes and mechanical construc-tion. An example of such a secondary resonance is schematicallydrawn in Fig. 3.8 with a dotted line (fx1, fx2).

The above gures precisely identify the few modes that exhibita resonance and that are actually excited by the given vibrationforce. There are more physical resonances than those exhibitedby the simulated resonance curves. The simulations are only roughapproximations of the geometry of the vibrating body. The realelastic constants are not linear and neither are the dissipationcoefcients.

The vibration amplitude at the tip of the transducer is calcu-lated between 10 and 40 kHz. Two maxima are evident, whichcorrespond to modes 6 and 15. These two modes are the purelongitudinal vibration modes of the transducer between 10 and40 kHz ( Fig. 3.9 ).

4. Experimental research

Because of the many variables that could potentially inuencethe consolidation of concrete, accurate and controlled measure-ments were made to prevent crediting the wrong variable for somemeasured response. In order to do so, experimental research wasset up and many experiments were carried out. A design of exper-iment (DOE) was performed to study the impact of blade distanceson both static friction tests and dynamic ow tests.

4.1. Construction of a valve model

In order to begin the experiments, a new valve model wasconstructed. The schematic of this valve is shown in Fig. 4.1 .

The horn in our system is the valve tip. The valve tip wasdesigned by machining several blades along the path of concrete(Fig. 4.2 ).

4.2. Measuring static friction

A method was devised for measuring the friction coefcient of the given viscous materials with the given planar valve bladesurfaces. To achieve this purpose different valves with varyingblade sizes were constructed as shown in Fig. 4.2 .

In this research both clay and fresh concrete were used assample materials. Concrete mass-per-volume for various mixes

Fig. 3.8. Frequency response of a piezoelectric transducer (phase and impedancealong Z axis). Source: Lorenzo [22] .

Fig. 3.9. Vibration amplitude in between 10 and 40 kHz. Source: Miodrag Prokic MPI Company.

Fig. 4.1. CAD models of the designed valve.



8/12

varied and was calculated from a measured sample. Thene aggre-gate used in this experiment consisted of different varieties includ-ing silica sand. There was also coarse aggregate in the concretemix, consisting of river gravel or crushed stone. The aggregategrain size was up to 5 mm ( Table 4.1 ).

Friction is inuenced by material viscosity. In order to measurethe viscosity the concrete was poured inside a cylindrical chamberrotating at constant speed. By measuring the variation of thetorque exerted on the shaft driving the chamber over time theviscosity of the concrete was derived. Although the measurementwas not entirely accurate, it was accurate enough to allow estima-tion of the range of material viscosity for the experiment. Thisidenties the friction coefcient used in our design procedure.

4.3. Factor identication and classication

Conducting detailed experiments for all identied factors re-quired many runs. However, by eliminating less important factorsthis research was conducted using fewer runs with a subset of fac-tors. Some of the factors found to signicantly impact the responsewere:

Pumping pressure of concrete. Size of blades along the direction of movement . Distance between blades. Amplitude and frequency of vibration. Viscosity of concrete. Proportion of sand. Grain size of sand. Density of concrete. On/off duty cycle.

In order to apply the best possible conguration, a properdesign of experiment was established. The most signicant vari-

ables affecting the outcome were selected as design variables.These design variables (DVs) are independent quantities that werevaried in order to achieve an optimal design. Upper and lowerlimits were specied to serve as constraints on the design vari-ables. These limits dened the range of variation for the DVs. Statevariables (SVs) are quantities that constrain the design. These were

also known as dependent variables and were typically responsequantities being functions of the design variables.

The objective function is the dependent variable that providesan overall indication of the valve performance. This function wasin terms of the DVs, that is, changing the values of the DVs changedthe value of the objective function which was the minimum vibra-tion energy required to overcome the friction and maintain theow.

4.4. Operational range determination

Before designing the experiments, the operational range foreach factor needed to be determined. For example, pumping pres-sure needed to be adjusted, concrete could be made more viscous,and the sand size and proportion of sand in the mix could be adjusted ( Table 4.2 ).

Determination of the maximum and minimum values for eachfactor was based on previous experience and a set of preliminaryplanned experiments. The preliminary experiments were basedon one-factor-at-a-time variation. However, this method mayunnecessarily eliminate part of the operational region for factorranges.

5. Conducting experiments

5.1. Static friction ow stop test

Here, several valves, with the same number and size of bladesspecied in the previous stage, were constructed. The valves dif-fered with respect to inter-blade distances. Then for each valve aset of experiments was conducted with different materials and dif-ferent pumping pressures. The data were input into a regressionmodel and an equation was derived to relate ow to friction sur-face size, concrete pressure, and material characteristics (includingviscosity and density). By putting the ow equal to zero, pressureto maximum, and having a standard material, the required surfaceto stop the ow in that condition was specied by computation.

The test stand shown in Fig. 5.1 was built in order to perform astatic test. Pressure was applied by a compressor to two pneumatic

cylinders. A one-inch diameter pipe was lled with one foot of con-crete material and pressure was incrementally adjusted between10 and 30 psi while the output volume per unit time was mea-sured. This way the ow rate at each incremental pressure wasdetermined.

5.2. Measuring ow with vibration

Vibratory friction reduction for viscous materials is achieved bysetting the particles into motion, thus overcoming static frictionbetween particles and between particles and inner surfaces of the conduit in which they travel. In order to accomplish this forconcrete valves the vibration test was conducted for each valvein the previously constructed set, with the vibration power set at

a certain value. Then for each valve conguration different pres-sure values and different materials were tried. The desired ow

Fig. 4.2. Valve-head varieties.

Table 4.1

Concrete viscosity with different aggregate size and water-to-concrete ratio. Source:Ferraris, C.

Classication Water/concreteratio

Plastic viscosity(Pa s)

Concrete with ne aggregate 0.55 41.2Concrete with course

aggregate

0.6 56.7

Table 4.2

Operating range for the experiment.

Parameter Min Operating point Max

Pressure of concrete 10 psi 15 psi 20 psiAmplitude of vibration 20% 50% 100%Viscosity of concrete 6000 Cst 11,000 Cst 16,000 CstGrain size of sand 2 mm 3.5 mm 5 mmOn/off duty cycle 0% 50% 100%

Density of concrete 2100kg/m 3 2300 kg/m 3 2500 kg/m 3

http://-/?-http://-/?-


9/12

rate was determined by the required pressure, provided that thetotal valve friction in the ON state was known. To estimate theON state friction an initial pressure (i.e., the pressure of the mate-rial at the inlet of the valve) was measured and used in the designprocess. The results were put into a regression model and an equa-tion was derived for the valve when in operation. Then by applyingthe maximum pressure in ON state the calculated friction surfacefrom the above the ow rate could be determined. If the ow ratewas in the desired range we followed the FEM analysis otherwisethe procedure was repeated. The rationale for repeating the pro-cess is that if the valve shape is designed solely based on ow itcannot be guaranteed that the natural frequency of that valvewould match the piezo element natural frequency, which is20 KHz. in that case tuning the valve shape would be requiredusing FEM modeling to assure the valve will resonate at the same

frequency as the piezo element. In some cases changing the valvedimension will affect the ow in which case all ow calculationsneed to be redone to assure that the valve can still deliver the sameamount of ow.

If we wanted a greater ow rate, then we needed greater pres-sure. In that case the surface size required to stop the ow wasgreater than the previous amount. This pressure, which was iden-tied experimentally in the design of experiments, was imple-mented after the design and construction of the valve ( Fig. 5.2 ).

5.3. Optimal valve shape

The challenge was to determine the number, size, and distancebetween the blades to be placed along the material ow path. The

optimal valve shape was then selected through a process that be-gan with an arbitrary starting design conguration that providedthe requisite total blade surface area. A heuristic search procedurewas then employed to converge toward the optimal (smallest withdesirable form factor) design. Fig. 5.3 shows some possible shapesof candidate valve designs.

The optimum valve shape is selected in such a way that maxi-mum vibration energy is transmitted to each surface and it guaran-tees that each surface receives enough vibration to dislodgematerial which is attached to its surface. If the attached materialcannot be moved after a certain elapsed time the material wouldcure inside the valve section and hence would render that portionof the valve useless. Usually longer dimension along the transduceraxis or movement direction give better results for which dissipa-

tion of energy is minimized. An example of a desirable shape isthe valve shown in Fig. 5.3 c, Also the valve should be designed in

such a way that prolonged vibration would not induce excessivefatigue in certain parts.

The distance between blades was decided to be greater than thelargest particle size. Subsequently the blades were designed insuch a way that the total surface would equal the result from theabove section.

The objective function in the optimization process was the totalactive displacement (i.e., movement caused by vibration) of blade

surfaces across the direction of ow. Other forms of vibration wereconsidered as dissipative vibrations that result in power loss.

6. Analytical research

FEM modeling was used at this stage to ensure that the reso-nance frequency of the designed valve would match that of thetransducerbooster delivering the vibration. The FEM model alsoprovided the total (integral) of all surface displacements as a singlenumber to be used for the response surface methodology (i.e.,objective function value for the given design).

6.1. Finite element analysis

Various numerical methods have been used to study the fre-quency characteristics and vibration modes for the coupled vibra-tion of piezoelectric sandwich ultrasonic transducers. Numericalmodels of high power ultrasonic systems areusually based on FEM.

It should be noted that due to the approximations in the FEMmodel certain less signicant vibration modes cannot be foundnumerically. This should cause no issues as in one-dimensionalsandwich transducer design theory it is assumed that such trans-ducers primarily vibrate in longitudinal mode and that radialvibration is negligible. This renders FEM a reliable analysis toolin this case. With respect to hardware design, this also means that

the lateral dimension of a transducer needs to be much less com-pared to its longitudinal dimension [13] .

The Computer Aided Resonator Design (CARD) software wasused to perform nite element analysis in this research. This soft-ware applies quantitative techniques to the design of ultrasonicresonators that vibrate in a longitudinal mode. CARD providesassistance in the design of resonators having low-to-moderatecomplexity [21] .

With CARD, alternative resonator designs were quickly evalu-ated hence the need for building hardware and physical testingwas dramatically reduced. The effects of several candidate resona-tors were easily determined using the software. CARD is especiallyuseful for designing low-stress resonators, resonators with a spec-ied gain, and resonators with a specied node location. The

following items were considered in the FEM analysis (based onthe guidelines given at the Krell Engineering site, 2002):

Fig. 5.1. Picture of the test stand.

Fig. 5.2. Model of a designed valve including transducer, booster and cover.

http://-/?-http://-/?-


10/12

Geometry : Insignicant changes in geometry can dramaticallyaffect the FEM predictions of resonator performance. The valvemodel geometry hence made a reasonably accurate representa-tion of the actual designed valve.

Material property : FEM cannot correctly predict the valveperformance unless the material properties, such as Youngsmodulus, modulus of rigidity, Poissons ratio, and density, areknown. In this test all values for materials properties weretaken from the available tables for generic material.

Frequencies : Frequencies are the easiest and most accurateparameter values to measure and verify. However, note that

FEM may predict more frequencies than can be measured inthe actual resonator. These extra FEM frequencies are oftenbending resonances or resonances at a node that runs throughthe resonators axis. Optimization on valve dimensions wasperformed in order to eliminate these unexpected frequencies.

Amplitudes : Amplitudes are relatively easy to verify, but caremust be taken to assure that the vibration amplitudes and theirlocations were correctly measured. This was especially impor-tant in this application as the amplitude readings could changerapidly depending on the sophistication of the amplitudemeasurement equipment. A 5% amplitude difference betweenthe measured values and the FEM values would not be unusual.Vibration amplitudes need to be measured at the tip of thetransducer and compared with the FEMresult. A large deviation

means that either calculation is wrong or the valve shape asmodeled is not accurate.

Stress : Stress usually cannot be directly veried because exper-imental stress data is usually not available, however, stressesyielded by FEM will not be correct if the FEM amplitudes arenot correct, because stress depends on the amplitude gradient.

Accuracy : The error in current methods for modeling ultrasonictransducers is on the order of 25% for frequency response and510% for amplitude response.

6.2. Design assumptions

Some of the design assumptions are as follows:

Mechanical components are assumed to vibrate longitudinallyonly. Hence, it is assumed that the vibration in this case isone-dimensional.

The propagation medium of the waves is assumed to be homo-geneous and that specic properties such as density, Youngsmodulus, etc., are constant.

The effect of elements such as bolts and electrodes on the sys-tem is negligible.

The transducer, the front and back materials and the piezoelec-tric cylinders have cylindrical shape.

7. Controlling ow with duty cycle

To achieve ow control a PWM (pulse width modulation)approach was used. The duty cycle of the pulse (i.e., the duration

Fig. 5.3. Possible shapes of candidate valve designs.

Fig. 7.1. 60% Duty cycle graph.

Fig. 7.2. 40% Duty cycle graph.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


11/12

of vibration ON divided by duration of the OFF state) inuences theoverall ow rate through the valve. The preference here was tomaximize the PWM frequency such that ow pulsation was mini-mized. Given that the uid has mass, which is subject to inertia,there was a limit on the deceleration time of concrete mass dueto the exerted friction force. At this stage two experimental activ-ities were performed for each vibration power to identify: (a) theduty cycle curve to achieve ow rates from zero to the specied(desired) maximum ow, and (b) the curve relating the PWMmaximum frequency to ow rate ranges ( Figs. 7.1 and 7.2 ).

The optimum duty cycle for each ow rate was the one that hadthe highest frequency (to minimize ow pulsation). Material massinertia would prevent stoppage if the duration of OFF cycles weretoo short. In other words, if OFF cycles were long enough theywould allow the material to reach the complete stop before thenext start.

Flow control with duty cycle primarily depends on the genera-tor response, and since in the case of the generator used in ourexperiments it takes about two seconds for the generator to startafter each stop, any duty cycle with an ON less than two seconds

did not measurably affect the ow rate.It was revealed that the material ow has linear relationship

with duty cycle. The slope of the chart in lower duty cycle is differ-ent from the higher duty cycle and the reason is mostly related togenerator response in lower duty cycles ( Fig. 7.3 ). By using a moreresponsive generator accuracy is increased and the curve canbecome more linear.

A programmable logic controller (PLC) was used at this stage tocontrol the frequency and duty cycle of the pulses to the generator.

8. Conclusion

In order to design a proper vibration-operated valve several

experiments were performed. First, the viscosity of candidatematerials was identied. Then a set of static pressure tests wereestablished and the ow equation relating to friction surface wasderived. The objective was to determine the surface size of theblades along the ow to provide the required maximum ow rateand stop the ow within a specic pumping pressure. By changingthe pressure and the material choice the required surface to stopthe ow was calculated using Response Surface Methodology(RSM). After performing the static test, a set of dynamic tests wereconducted by implementing the piezo vibration. The pressure wasadjusted and ow at each pressure was determined. If the pressureand ow were within range, then the valve designs were optimizedusing a nite element analysis approach.

Vibration operated-valveshave certain unmatched advantages in

comparison with other valve solutions for cementitious materialow control. Most notably, such valvesdo nothave mechanical mov-

ing parts that could jam and they lack rubber components that areprone to rapid wear. Other valves used for ow control of uids withabrasive particles are bound to use rubber and hence their useful lifeis very limited. Another drawback of conventional valves is theinability to perform proportionalow control dueto thegrain-bridg-ing phenomenon which could completely shut down the ow (dueto clogging) in partially open valve states. Such problems make itimpossible to use the conventionalvalves in continuous ow controlmechanisms such as vibration-basedcongurations.Vibrationbasedvalves seem to be ideal candidates for such applications.

Acknowledgement

The National Science Foundation, the Ofce of Naval Researchand the Annenberg Foundation have partially supported the re-search reported in this paper.

References

[1] ACI committee report: behavior of fresh concrete during vibration. ACI, ReportNo. 309 1R-81.

[6] Khoshnevis B. Automated construction by contour crafting related robotics

and information technologies. J Automat Constr 2004;13(1):519. January .[8] Ferraris C. Relating fresh concrete viscosity measurements from different

rheometers. J Res Natl Inst Stand Technol 2003;108(3). MayJune .[9] Ferraris C. Fresh concrete rheology: recent developments. Building and re

research laboratory. American ceramic society. Materials Science of ConcreteVI, Sidney Mindess and Jan Skalny; 2001. p. 21541.

[10] Ferraris C. Measurement of the rheological properties of high performanceconcrete. State Art Rep 1999;104(5). SeptemberOctober .

[11] Ferraris C. Simulation of SCCow. < http://ciks.cbt.nist.gov/~garbocz/scc_2003/Simulation.SCC.htm >; 2002.

[13] Ando E. Finite element simulation of transient heat response in ultrasonictransducer. IEEE Trans Ultrason, Ferroelectr, Freq Cont 1992;39(3). May .

[15] Loadwick F. Some factors affecting the ow of concrete through pipelines. In:First international conference on the hydraulic transport of solids in pipes;1970.

[21] Krell engineering. Ultrasonic resonator design using nite elementanalysis (FEA)..

[22] Parrini L. Design of advanced ultrasonic transducers for welding devices. WeldDev Ultrason, Ferroelectr, Freq Cont 2001;48(6). November .

[27] Neppiras EA. The pre-stressed piezoelectric sandwich transducer. Ultrason IntConf Proc 1973:295302 .

Further reading

[2] Khoshnevis B, Behdinan K. Vibration operated valve and ow metering devicefor abrasive viscous uids. In: ISA 58th international instrumentationsymposium; 2012.

[3] Behdinan K, Khoshnevis B. Methodology for design of a vibration operatedvalve for abrasive viscous uids. In: ISA 57th international instrumentationsymposium; 2011.

[4] Asiabanpour B, Palmer K, Khoshnevis B. An experimental study of surfacequality and dimensional accuracy for selective inhibition of sintering. RapidPrototyp J 2004;10(3):18192 .

[5] Asiabanpour B, Palmer K, Khoshnevis B. Performance factors in the selectiveinhibition of sintering process. In: Institute of industrial engineering researchconference. Portland, OR; 2003.

[7] Ferraris C. Testing and modeling of fresh concrete rheology. Natl Inst StandTechnol 1998. February .[12] Dion J, Cornieles E, Galindo F, Kodjo A. Exact one dimensional computation of

ultrasonic transducers with several piezoelectric elements and passive layersusing the transmission line analogy. IEEE Trans Ultrason Ferroelectr1997:112031. Freq. Control 44 .

[14] Lim F, Cardoni A, Lucas M. Optimization of the vibration response of ultrasoniccutting systems. IMA J Appl Mathemat 2005 .

[16] Frederick JR. Ultrasonics engineering. New York: Wiley; 1965 .[17] Murata J, Kikukawa H. Viscosity equation for fresh concrete. ACI Mater J 1992.

MayJune, Title No. 89-M24 .[18] Assaad J. Effect of coarse aggregate characteristics on lateral pressure exerted

by self-consolidating concrete. Am Concr Inst 2005. May/June .[19] Kolek J. Research on the vibration of fresh concrete. US Army Engineer

Waterways Experiment; 1958 .[20] Khmelev V, Slivin A, Barsukov R, Tsyganok S, Savin I, Shalunov A. Levin S,

Abramov V. Measurement of parameters and automatic selection of optimalmodes during ultrasonic welding of thermoplastic materials. In: 7th IntSiberian workshop and tutorials (Erlagol, July 2006) session VI; 2006.

[23] Prokic M. Piezoelectric transducers modeling and characterization. MPISwitzerland; 2004.

Fig. 7.3. Flow versus duty cycle graph.

http://refhub.elsevier.com/S0957-4158(13)00125-6/h0010http://refhub.elsevier.com/S0957-4158(13)00125-6/h0010http://refhub.elsevier.com/S0957-4158(13)00125-6/h0010http://refhub.elsevier.com/S0957-4158(13)00125-6/h0020http://refhub.elsevier.com/S0957-4158(13)00125-6/h0020http://refhub.elsevier.com/S0957-4158(13)00125-6/h0025http://refhub.elsevier.com/S0957-4158(13)00125-6/h0025http://ciks.cbt.nist.gov/~garbocz/scc_2003/Simulation.SCC.htmhttp://ciks.cbt.nist.gov/~garbocz/scc_2003/Simulation.SCC.htmhttp://refhub.elsevier.com/S0957-4158(13)00125-6/h0035http://refhub.elsevier.com/S0957-4158(13)00125-6/h0035http://www.krell-engineering.com/fea/fea_info/fea_resonator_design.htmhttp://refhub.elsevier.com/S0957-4158(13)00125-6/h0070http://refhub.elsevier.com/S0957-4158(13)00125-6/h0070http://refhub.elsevier.com/S0957-4158(13)00125-6/h0070http://refhub.elsevier.com/S0957-4158(13)00125-6/h0085http://refhub.elsevier.com/S0957-4158(13)00125-6/h0085http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0015http://refhub.elsevier.com/S0957-4158(13)00125-6/h0015http://refhub.elsevier.com/S0957-4158(13)00125-6/h0015http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0040http://refhub.elsevier.com/S0957-4158(13)00125-6/h0040http://refhub.elsevier.com/S0957-4158(13)00125-6/h0050http://refhub.elsevier.com/S0957-4158(13)00125-6/h0055http://refhub.elsevier.com/S0957-4158(13)00125-6/h0055http://refhub.elsevier.com/S0957-4158(13)00125-6/h0055http://refhub.elsevier.com/S0957-4158(13)00125-6/h0060http://refhub.elsevier.com/S0957-4158(13)00125-6/h0060http://refhub.elsevier.com/S0957-4158(13)00125-6/h0065http://refhub.elsevier.com/S0957-4158(13)00125-6/h0065http://-/?-http://refhub.elsevier.com/S0957-4158(13)00125-6/h0065http://refhub.elsevier.com/S0957-4158(13)00125-6/h0065http://refhub.elsevier.com/S0957-4158(13)00125-6/h0060http://refhub.elsevier.com/S0957-4158(13)00125-6/h0060http://refhub.elsevier.com/S0957-4158(13)00125-6/h0055http://refhub.elsevier.com/S0957-4158(13)00125-6/h0055http://refhub.elsevier.com/S0957-4158(13)00125-6/h0050http://refhub.elsevier.com/S0957-4158(13)00125-6/h0040http://refhub.elsevier.com/S0957-4158(13)00125-6/h0040http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0030http://refhub.elsevier.com/S0957-4158(13)00125-6/h0015http://refhub.elsevier.com/S0957-4158(13)00125-6/h0015http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0005http://refhub.elsevier.com/S0957-4158(13)00125-6/h0085http://refhub.elsevier.com/S0957-4158(13)00125-6/h0085http://refhub.elsevier.com/S0957-4158(13)00125-6/h0070http://refhub.elsevier.com/S0957-4158(13)00125-6/h0070http://www.krell-engineering.com/fea/fea_info/fea_resonator_design.htmhttp://refhub.elsevier.com/S0957-4158(13)00125-6/h0035http://refhub.elsevier.com/S0957-4158(13)00125-6/h0035http://ciks.cbt.nist.gov/~garbocz/scc_2003/Simulation.SCC.htmhttp://ciks.cbt.nist.gov/~garbocz/scc_2003/Simulation.SCC.htmhttp://refhub.elsevier.com/S0957-4158(13)00125-6/h0025http://refhub.elsevier.com/S0957-4158(13)00125-6/h0025http://refhub.elsevier.com/S0957-4158(13)00125-6/h0020http://refhub.elsevier.com/S0957-4158(13)00125-6/h0020http://refhub.elsevier.com/S0957-4158(13)00125-6/h0010http://refhub.elsevier.com/S0957-4158(13)00125-6/h0010http://-/?-


12/12

[24] Murat Guney, Es. Modeling of multilayered piezoelectric transducers withultrasonic welding application. Smart Mater Struct 2007;16(2007):54154 .

[25] Badescu Mirecea, Bao Xiaoqi, Bar-Cohen Yoseph, Chang Zensheu, SherritStewart. Integrated modeling of the ultrasonic drill procedure and analysis

result. In: Proceedings of the SPIE smart structures conference. San Diego, CA.,SPIE Vol. 576437, March 710; 2005.

[26] National institute of standard and technology. Concrete ow simulation..

http://refhub.elsevier.com/S0957-4158(13)00125-6/h0080http://refhub.elsevier.com/S0957-4158(13)00125-6/h0080http://www.nist.gov/public_affairs/techbeat/concrete.htmhttp://www.nist.gov/public_affairs/techbeat/concrete.htmhttp://refhub.elsevier.com/S0957-4158(13)00125-6/h0080http://refhub.elsevier.com/S0957-4158(13)00125-6/h0080

methodology for design of a vibration operated valve for abrasive

Documents