Automation in Construction 20 (2011) 11331142

Contents lists available at ScienceDirect

Automation in Construction
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / a u t c o n

A test rig and numerical model for investigating truck mounted concrete pumps
G. Cazzulani, C. Ghielmetti, H. Giberti, F. Resta, F. Ripamonti
Mechanical Engineering Department, Politecnico di Milano, Via La Masa, 1, 20186 Milano, Italy

a r t i c l e

i n f o

a b s t r a c t
Concrete pump booms are subjected to vibrations that increase mechanical stress and shorten their lifespan. This paper aims to study the problem by considering the two subsystems, the boom and the concrete pump, that have the greatest effect on the phenomenon. The authors supply numerical and experimental tools that can analyze the problem in depth in all its complexity. First, the systems were investigated independently, to identify their individual aspects. Then a mathematical model was created to reproduce the behavior of the whole system and the interaction between boom and pump. The result was a new powerful tool for investigating passive and active solutions for suppressing vibration. 2011 Elsevier B.V. All rights reserved.

Article history: Accepted 5 April 2011 Available online 10 May 2011 Keywords: Nonlinear multibody dynamics co-simulation Concrete pumping Flexible system

1. Introduction In recent years construction machinery has undergone considerable development, in terms of increasing dimensions and reduced weight. This trend can be clearly seen in, for example, concrete pump booms. The need for weight reduction, to improve the performance of the systems, makes for structures characterized by high exibility and low damping. These characteristics, coupled with the dynamic loads due to the large motion of the structure and the ow of the concrete, produce considerable mechanical stress on the material. These structures suffer from fatigue and instability issues, with several negative implications for safety. To reduce the spreading of cracks and increase the booms' working life, it is necessary to reduce the stress by suppressing vibrations of the entire system. Traditional external passive control methods are generally more invasive, since they involve introducing mass on to the structure, and are less effective in a large range of frequencies. On the other hand, active control is an attractive solution, especially considering the rapid development of computer hardware and the consequent reduction of costs. In both cases, a numerical tool that can reproduce the dynamics of the entire system is very useful for identifying the best solution for each specic model and specic application. Some contributions can be found in the literature. For example, Khulief [1] used an FEM model of a exible boom to dene a control logic for suppressing vibration. Other authors created boom models for tip trajectory synthesis (Wang [2]) or the automation of the pouring process (Zhou [3]) and pump models for enhanced component design (Worthington [4]). Finally the papers on innovative long span machines by Kronenberg [5] [6], specically relating to

the Putzmeister AG company, should be mentioned, though they contain no analysis of possible control laws. The aim of this paper is to investigate and reproduce the dynamics of the entire system, which have been little investigated in the scientic literature, creating a numerical and experimental development environment for investigating solutions aimed at increasing the performance of the system in terms of pumping capacity, vibration and safety, highlighting the most critical aspects of the system as a whole. The complexity of the system led us rst to consider the boom and the pump independently. Two test rigs were built to reproduce in the laboratory the characteristics of the systems and permit an intensive series of experiments. The results of these experiments were the starting point for dening the mathematical models. For the boom, a non-linear exible multibody (MB) model was created, while the pumping group numerical model was used to solve the oil and concrete continuity equations and the equations of motion of the hydraulic pistons. Then, in the second stage, the forces coming from the pump model were used as the input for the boom model, that, on the other side, returns the system conguration. 2. The truck mounted pump Concrete pump booms are complex dynamic systems with a variable number of links linked together by kinematic joints and moved by hydraulic actuators. In this way the end-activator, i.e. the boom tip housing the device for placing the concrete, may be very far from the pump (Fig. 1-a) or in very high place (Fig. 1-b). In general these systems have high exibility and, as a result, very low natural frequencies. Therefore, because of the very low damping associated with these frequencies, there will be high amplitudes of vibration and severe mechanical stresses on the structure. Moreover, it is important to underline that the possibility of changing the conguration of the system spreads the natural frequencies over a wide range. This range can be only estimated in the design

Corresponding author. E-mail address: francesco.ripamonti@polimi.it (F. Ripamonti). 0926-5805/$ see front matter 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.autcon.2011.04.015

1134

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

Fig. 1. The system: the long span (left) and the high place (right) congurations.

stage and does not exclude resonance forcing during the boom's lifespan. In fact, in addition to the forcing associated with the boom's motion that excites the system's vibration modes, during normal operation the structure is affected by the pumping of the concrete. The pumping unit is bolted to the chassis of the truck (Fig. 2) and, through its connection to the truck's engine, works like an alternating pump, consisting of two actuators and four pressure chambers. The alternative motion of the pistons forces the concrete to ow into a pipe connected to the boom; this is clearly visible in Fig. 2. The pistons movement, varying the pressure in the chambers, generates forces on the boom. In particular, the pumping forces are due to the concrete owing through the pipe that is rigidly connected to the boom segments. The concrete mainly transfers the forces due to the friction and the pressure difference between the inlet and outlet surfaces. The sum of these contributions denes a longitudinal component along the pipeline, balanced by the reaction force of each single link (Fig. 3-a). These reactions are perceived by nonaligned segments as transverse forcing in the vertical plane (Fig. 3-b). In conclusion, as said before, boom vibrations are mainly due to two contributions: the boom large motion exciting its natural frequencies, and the concrete owing forcing the boom at the pumping frequency. Fig. 4 shows an example of these two situations in terms of acceleration of the nal segment tip. Fig. 4-a shows the decay of the acceleration after movement ends; the harmonic component at the rst natural frequency (0.47 Hz for the long span conguration) is clearly visible. Fig. 4-b

shows the steady state response (in terms of acceleration) due to the pumping of the concrete; we can see that the fundamental forcing component (0.45 Hz) is very close to the system's rst natural frequency in the long span conguration. For all these reasons, and in particular because of the complexity involved in studying the system as a whole, this paper will consider, initially, the boom and the pump independently. This will allow the most important parameters of both subsystems to be highlighted and the best solutions for improving performance and safety identied. In particular, by considering the pump and the boom respectively as cause and effect, it is possible to reduce oscillations both by improving the pump design and/or by applying a control logic for suppressing vibration. Below two experimental test rigs are described: a 1:3 scale model of the boom and a full scale test rig for the pump system. 3. The boom test rig This section describes the scaled-down test rig. The aim was to obtain a reduced-scale system that reproduced the same dynamic behavior of full-scale booms. For this reason, the most important characteristics for the test rig were: natural frequencies: geometric, inertial and elastic characteristics needed to be chosen so that the rst natural frequencies and modal

Fig. 2. The truck concrete pump boom (left) and the pumping unit (right).

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

1135

Fig. 3. Diagram of the pumping forcing the link/pipe interaction (left) and the forcing on the entire system (right).

shapes were comparable with those of full-scale booms. In fact, the high exibility (and as a consequence the low natural frequencies) associated with low structural damping is the main cause of highamplitude vibrations. Table 1 shows the frequency range imposed, considering the boom in horizontal conguration and assuming rigid actuators.

available working congurations: the test-rig must be able to reproduce the typical working congurations of the real booms, to simulate boom behavior in real working conditions. For this reason, to obtain a large number of congurations, the actuator kinematics were dimensioned so as to obtain the largest link rotations while keeping the actuator forces within their limits. Other constraints are the maximum total boom length (xed at 12 m) and actuator typologies. There are two typologies of actuators available, called simply A and B below (their main characteristics are given in Table 2). Both kinds of actuators are moved through a position PID controller that simulates both the behavior of the human operator during boom movement and the blocking valve, that maintains the boom position when the operator does not move the boom. In deciding on the number of links in the boom test rig, some observations must be made. On the one hand, a better simulation of full-scale booms can be created by using a test rig with several links. In fact, large-dimension booms (on which vibrations are usually more critical) have a large number of links (typically between 5 and 7). On the other hand, once actuator dimensions and total boom length are xed, a greater number of links (and, as a result, of actuators) make the structure stiffer, and move the boom's natural frequencies out of the desired range. These considerations led to the construction of a three-link boom (Fig. 5), whose main characteristics are given in Table 3. The links have decreasing sizes, like those of full-scale booms. The rst and second link are moved using A type actuators, the third using the B type (see Table 2). Concrete being pumped along the boom can be simulated using additional weights (Fig. 6-a) and applying external forces to the structure. The weights simulate the additional inertial contribution introduced by the concrete, while the forces, applied through external actuators (Fig. 6-b), simulate the stress due to the ow of the concrete. The test rig was tted with several sensors. Each actuator had an LVDT sensor and a load cell. The LVDT sensor measurements can be used both by the PID controller and for calculating link angles and hence the boom conguration. The load cell measures the total force applied by the actuators to the boom.
Table 1 Imposed range for the rst and second natural frequencies. Imposed range

Fig. 4. The system forcing: boom tip acceleration due to movement (a) and to pumping in long span conguration (b).

1st natural frequency 2nd natural frequency

0.50 0.80 1.50 2.00

1136 Table 2 Characteristics of the hydraulic actuators available. Max load (+/-) [kN] Model A Model B 50 15

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

Nominal length [mm] 1320 1460

Stroke (+/) [mm] 125 125

Finally three capacitive accelerometers, positioned on each link tip, were used to measure the vibrations in the boom. Several tests were carried out on the rig to identify its dynamic behavior. As the system is non-linear, its parameters depend on boom conguration. Once the boom position was xed, its behavior can be assumed linear and natural frequencies and damping ratios can be estimated. Since both natural frequencies and damping depend on the conguration, the estimation tests were repeated for different positions. The experimental characterization was performed using different tests: actuator static force measurement, which was used to verify the model of the inertial contributions of the boom; natural frequencies identication, performed using chirp signal tests; estimation of the damping factor associated with each mode, carried out by exciting the structure on its natural frequencies and considering the decay time histories. The rst two (static actuator forces and natural frequencies) were used to verify the precision of the mathematical model of the boom that will be described in Section 5. Modal damping was used to set the damping parameters of the model (refer to Section 5). 4. The concrete pumping unit test rig In this section we will describe the pumping unit layout. The test rig and all its auxiliary circuits were built to reproduce the actual working pump and were used to identify its parameters. Fig. 7 shows the layout of the pumping unit test rig; the positioning of the unit, the way it is xed to the ground, its inclination and all circuits provide a perfect simulation of the real unit mounted on the truck. The test rig is composed of the pumping unit (2 in Fig. 7) and of the auxiliary circuits: the hydraulic circuit and the concrete circuit (called the CLS circuit). It works in this way: the work uid ows out from the cylinders of the pumping unit and returns to its tank after passing through a closed circuit consisting of a pipeline, a series of regulating valves and a steel holding tank almost three meters high (1 in Fig. 7). The oil pumps are powered by diesel engines whose rpm can be varied to regulate the input oil ow (3 in Fig. 7). The pressure in the chambers depends on the load created by the concrete being pumped. This load is a function of two parameters: piston velocity and the position of the boom. In the rst case there is direct proportion to the velocity of the pistons because of the resistance developed by the concrete (friction), while in the second case the height of the boom creates the static pressure of the column. For this reason the layout

was designed to simulate such conditions. Both the friction force and the static pressure are simulated through a dedicated circuit. In particular this circuit can simulate the pressure of concrete by adjusting the position of certain valves. This complete test rig makes it possible to carry out many kinds of experiments. Fig. 8 shows a section of the mechanical system. It describes its components and shows how they interact with each other to pump the concrete along the metal pipeline. The pumping unit is a volumetric pump with two pistons (E), that alternately suck up concrete from the concrete tank (H) and pump it out along the pipeline (I). The pistons are driven by two hydraulic actuators (B and C). To direct the pumped ow towards the outlet pipeline a device called an S Valve (G) is used. It is possible to divide the pumping unit system into two subsystems: the oil side and concrete side. The oil side represents the power unit while the concrete side is the driven unit and is in direct contact with the concrete. The pumping unit works cyclically and each cycle is composed of two phases: ejection and replenishment. In the rst phase one of the two actuators (C for example) pushes out the concrete accumulated in the pipes (see outgoing concrete in Fig. 8) while the second actuator (B) is driven by the rst one through a hydraulic circuit called Slave (D) and moves in the opposite direction almost simultaneously, relling its cylinder with concrete (see ingoing concrete in Fig. 8). The elements making up the oil side are the oil pump (P), the ducts (A), the proximity sensors (X) and the slave. The pump moves the pistons, sending a ow of oil into one of the two chambers next to it. At the same time, an equal ow of oil returns to the pump from the other chamber. The oil works alternately in a closed loop formed by the pump and by the chambers C1 and C4. The pistons' alternating motion is controlled by two proximity sensors. These are intercepted by the passive piston when it is almost at the end of its stroke, and send an electric signal to invert the oil pump and the position of the S valve. Fig. 8 also shows a series of pipes called ducts, with retaining valves, which form a closed circuit on the oil side cylinder. They function as brakes, as oil conduits (because of oil leakage) and as a piston interlock during the operating cycles. These components are very important for the trade-off of pressure and ow between the adjacent chambers C1 and C2 (or C4 and C3). The concrete side elements are the S valve and the concrete tank. The S valve connects the end of the cylinder in the concrete side subsystem with the beginning of the pipeline. The rotation of this valve is, as mentioned above, triggered by a signal from the proximity sensors. The concrete tank is continuously relled by a concrete mixer and holds the uid used during the cycles. Beyond the physical components, the test rig is equipped with sensors for measuring outow and friction forces. Each chamber is tted with a pressure sensor and the actuators with a magnetostrictive rod for measuring their displacement. To measure the input ow of oil and the output ow of concrete, ow sensors were used. All these instruments were necessary and used both for direct and derived measurements. Some measurements, for example the velocity and acceleration of the pistons, were extracted from the measurements of displacement and numerically calculated. In addition, to simulate the static load of the concrete due to the conguration of the boom, a series of globe valves are installed. 5. Mathematical models and simulations As mentioned earlier, the mathematical models of the two subsystems were created using the data collected on the test rigs. The causeeffect relationship between the pump and the boom, with respect to the oscillation phenomenon during concrete pouring, allowed them to be treated separately in the mathematical model too. The models of each subsystem were calibrated independently and the combined simulation then implemented. In the cosimulation, the output of each subsystem is passed to the other as an input.

Table 3 Link dimensions. Length [m] Link 1 Link 2 Link 3 4.5 4.0 3.5 Mass per meter [kg/m] 18 15 10

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

1137

Fig. 5. The test rig (left) and a diagram of it (right).

5.1. The mathematical model of the boom The mathematical model of the boom was developed to simulate both its large motion and the vibrations due to its exible nature. As the boom is perfectly symmetrical with respect to the vertical plane containing the boom and its motion is planar, the system can be considered two-dimensional. This allows the model to be considerably simplied (lateral and torsion vibrations are neglected) without compromising its accuracy. Other assumptions are small deformations: the relationship between strain and deformation can be considered linear; in addition axial and bending deformations can be considered as decoupled. small links rotation speed: this assumption means we can neglect the contribution of centrifugal and Coriolis terms in the motion equation

Each link movement is described, using the so-called oating frame of reference formulation[7-9], by two sets of coordinates, describing respectively the large motion and the vibrations. The rst set (Fig. 9-a) describes the rigid motion of the jth link, in terms of its absolute displacement and rotation, considering the motion of the reference system of this link (y jO jx j) with respect to the global reference system (yOx). The translation contribution can be written as a function of the degrees of freedom of the previous links, while the rotation contribution is described by the independent variable j. The second set of coordinates (Fig. 9-b) models the link vibration with the nite element method (FEM), using beam elements. The deformations of each element are described considering the displacement and rotation of the nodal reference system (y jkO jkx jk) with respect to the link reference system (y jO jx j). Thanks to the assumption of planar motion, each FEM node is described by a vector, d jk, with three independent coordinates.

Fig. 6. Additional weights on the third link (left) and the external actuator simulating the pumping forces (right).

1138

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

Fig. 7. Complete pumping unit test rig layout (left) and a diagrammatic representation (right).

Dening the vector containing all the link rotations as and as d the vector containing all the nodal coordinates, the total independent variable vector (z) can be written as: z= & ' d 1

As said, Eq. (2) can be linearized about a conguration z = zj. Considering small deformations, the linearization conguration is dened as & zj = j 0 ' 3

Using the Lagrange formulation, we obtain the non-linear equation of the boom: Mzz = f z;z + where M(z) represents the inertial contribution of the structure (depending on boom position); f(z, z ) contains the elastic and damping terms; act(z) represents the kinematic relationship between the movement of the actuators and the independent coordinates, while Fact(t) is the actuator force vector; F(t) contains all the external forces acting on the boom. This nonlinear equation can be integrated, obtaining the numerical solution under generic actuators movement and external forces. Anyway, to easily compare it with the real system, it can be useful to consider some xed boom congurations. In this situation the boom behavior can be assumed linear and the nonlinear equation can be linearized into the well-known 2nd order formulation typical of a linear mechanical system. This process allows to compare the model and the real system in terms of damping and natural frequencies.
T act zFact t

and the linearized equation becomes   T j + Kj zj = f g zj + j Fact + Ft Mj zj + Rj z 4

+ Ft

where zj is dened as zj = z zj and describes the system's vibration. The damping matrix Rj was assumed to be the sum of two contributions: structural distributed damping Rj,str, expressed by Eq. (5) as a linear combination of the inertial and elastic terms Mj and Kj; as described in Section 3, the two coefcients and have been calculated starting from the experimental estimation of the damping associated to the boom modes. lumped damping Rj,conc described by three torsional dampers on the joints connecting the links. Rj;str = Mj + Kj 5

The damping parameters (, and the lumped terms) were computed minimizing the difference between estimated damping and model damping in different congurations.

Fig. 8. A detailed diagram of the pumping group.

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

1139

The state variable vector x is composed by the chamber pressures Pi and the pistons displacement xi, while the measurements vector yoil contains the oil uxes between the chambers Q ij. x= P1 P2 P3 Q23 P4 Q34 1 x 2 x Q 4PT g
T

x1

x2

7 8

yoil = f Q12

For the pressures in the oil side chambers, the following general equation is used   dV ! = Q ij Q jk i P i Vi dt 9

where is the oil bulk modulus and the Vi is the chamber volume, a function of yi. The oil ows can be evaluated using the well known general algebraic efux equation. In addition to these equations, the state of the system is completed by the equations of motion of the two pistons. These describe the motion of the system and couple the oil and concrete sides. They are summarized as 8 1 + mg sin P1 A1 P2 A2 Pcls6 Acls6 f x > > x = > > 1 m > > > <x 1 1 = x > 2 + mg sin > P A P3 A3 Pcls7 Acls7 f x > = 4 4 > x2 > m > > : 2 = x 2 x

10

The balance of forces is shown in Fig. 12, using piston 1 as an example. It takes into account the pressure of the oil and concrete chambers, the forces of and friction and the inertial force. gravity 1 is calculated with the LuGre friction force The friction force f x model [11] for translational friction. This model is generally used for hydraulic actuators and is given as
Fig. 9. Link reference systems (a) and nite element reference systems (b).

= Fc + Fs + Fv sign x f x

11

This numerical model can simulate the typical operating condition of concrete pump booms. As an example, Fig. 10 shows a numerical/ experimental comparison in terms of transfer function between input and tip acceleration. The input is a sweep-sine function between 0 and 7 Hz, obtained as an imposed movement of the rst actuator. The comparison shows good agreement between numerical and experimental results. The correlation is fairly good and it worsens increasing the frequency. This situation is quite common for mechanical systems, where the parameter estimation is easy for the low frequency modes, while it becomes more difcult for the higher frequency ones. For this application this result is acceptable and the numerical model developed can be used to study the dynamic behavior of booms and to develop control logics to suppress system vibrations. 5.2. The mathematical model of the pumping unit The pumping unit subsystem [10] was modeled with differential equations describing both the dynamics of oil chamber pressures and the motion of the pistons. The rst part of the system (oil side) was modeled as a series of chambers in pressure coupled by the oil ows Q ij. The model, represented in Fig. 11, is characterized by the input variables Q pump-i and PT, that represent respectively the pump ux and the oil tank pressure. & uoil = Q pumpi PT ' 6

and is assumed to be the It is a function of the relative velocity x sum of Stribeck, Coulomb and viscous (Fig. 13). components j Stribeck friction, Fs = Fbrk Fc e c x j x , is a decreasing friction and the negatively sloped characteristics are signicant at low velocities.

Fig. 10. Numerical/experimental comparison of the transfer function amplitude (a) and phase (b).

1140

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

Coulomb friction, Fc, is a constant force at any velocity. Viscous , gives a force directly proportional to the relative friction, Fv = r x velocity. The sum of Coulomb and Stribeck friction, at velocity close to zero, is often referred to as breakaway friction, Fbrk. Finally the concrete side (called the CLS side below) is governed by algebraic equations based on an energetic approach. The input variable, contained in the ucls vector, is & ucls = PTcls h ' 12

agreement is evident, and this was the starting point for the cosimulation described below between the pumping unit and boom environments. 5.3. The boom/pump interaction model The two described models and test rigs are able to reproduce a part of the whole behavior of a real boom. The aim of the nal part of the work was to simulate the complete behavior of a full-scale commercial boom in working conditions. This operation has been performed implementing a co-simulation between the two models. In particular, the boom one has been calibrated on the parameters of a full-scale boom. Fig. 16 shows the interaction model; the external inputs are the strokes of the actuator, which allow the boom conguration to be calculated. The information about the boom position and dynamics is passed to the pump model, which calculates the needed pump head. On the other side, the pump model will give the ow rate of the concrete and the pressure along the pipe, which can be used to calculate the pumping forces on the boom. In fact, as a rst approximation, the equilibrium of a generic concrete volume V is described by V v dV = Ain Pin dAin Aout Pout dAout Fcls t 15

where PTcls represents the pressure of the concrete tank (typically equal to the atmosphere pressure) and his the maximum height of the boom needed to determine the pressure of the CLS side chamber as Pclsi = cls gh + H x 13

The term H is the sum of concentrated and distributed losses and is a function of a concrete ow. The pressures in the CLS chambers (C6 and C7) and the output ow of concrete dene the CLS measurement vector 8 9 < Q cls = = Pcls6 : ; Pcls7

ycls

14

A6 ( is the coefcient of replenishment) and the in which Q cls = x pressures are alternately equal to Patm and Pclsi. We can then obtain the pressure of concrete at every point in the pipeline by applying Eq. (11). The complete system can be described in state-space form combining the oil side and CLS side equations. Fig. 14 shows the result of a pumping simulation in terms of state variables. In this case the test has a constant frequency of 0.25 Hz and overpressure of 40 bar: On the pumping unit system, a sensitivity analysis, involving the values of oil ow and globe valve area (simulating different h values) was carried out. The tests carried out yielded important information on the forces transferred from the pumping group to the boom; these were mainly due to CLS pressure in the pipeline during the work cycle. As validation of the numerical model, Fig. 15 shows a comparison with the experimental data, in terms of CLS ow and pressure. Good

where Fcls is the friction force that the concrete transfers to the boom and v/t represents the acceleration of the concrete volume. The acceleration vector is the sum of the acceleration due to the boom motion and the one due to the concrete ow along the pipe. As for the boom motion, the Coriolis and centrifugal terms have been neglected since the links rotation speed is low. Pin and Pout represent respectively the pressures on the input and output surfaces of the concrete volume. The concrete pumping force is transmitted to the boom through several links (Fig. 3-a) along the length of the pipe. Assuming constant pipe section and incompressible uid, the friction force in a nite volume can be calculated by integrating Eq. (15) as v + Pin Pout A t

Fcls = V

16

Fig. 11. Diagram of the pumping unit oil side.

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

1141

Fig. 12. Balance of the piston forces in the pumping unit.

This force acts on the boom model during motion integration as an external disturbance force. Fig. 17 shows the result of a co-simulation using the boom/pump interaction model in long span conguration. For a comparison with experimental data (Fig. 4), the simulation has been performed in the same conditions (long span conguration in Fig. 1 with a pumping frequency of 0.45 Hz). The comparison of boom tip acceleration shows a good agreement between numerical and experimental data. The main differences between the numerical and experimental data are related to the variability in the boom behavior between different cycles that are not reproduced by the numerical simulation (see for example the part of the history around 15 s). Anyway this variability is quite low and, for this reason, the complete boom/pump model can be considered as a tool able to reproduce the behavior of the entire system. 6. Conclusions The article describes a new experimental test rig created to study vibration control for a concrete pump truck boom. The experimental setup is split into two independent test rigs. The rst is a scaled model of the boom with three actuated links, instrumented with accelerometers to evaluate the vibratory state. The use of servo-actuated pistons allows to reproduce the system large motion with high accuracy and repeatability. The pumping effect due to the concrete owing has been reproduced with additional masses and external forces. The second is a concrete pump tted with instrumentation and connected to purpose-built hydraulic test circuit. This full scale test rig allows to investigate the internal pressure and the ow rate in the concrete and oil circuits. This article also presents mathematical models for describing the behavior of the physical systems. The mathematical models are validated and implemented in a software which simulates system behavior under varying operating conditions. This experimental setup and mathematical model are unique. It is a development environment that makes it possible to improve the performance of the concrete pump truck boom in terms of pumping capacity, maximum pumping distance attainable and safety.

In summary, the main goals achieved by this development environment are: 1. the possibility of developing and designing new boom position control strategies and new controls for pump ow. This result can be obtained simulating the boom and pump behavior separately, or together using the co-simulation tool. 2. the possibility of checking the control strategies developed on physical models with full instrumentation (the test rig). The test rig accurately reproduces the behavior of a concrete pump truck boom but in a completely safe and protected environment. 3. the possibility of verifying the robustness and dependability of the control strategies developed. In fact with this test rig it is easy to generate operating conditions difcult to reproduce in real systems and in real environments. 4. the ease of transferring the control strategies developed to real equipment.

Fig. 13. Translational friction model.

Fig. 14. Numerical simulations state variables; oil side pressures and piston speed.

1142

G. Cazzulani et al. / Automation in Construction 20 (2011) 11331142

Fig. 17. Co-simulation results: acceleration of the boom tip in long span conguration with a pumping frequency of 0.45 Hz.

References
[1] Y.A. Khulief, Vibration suppression in rotating beams using active modal control, J Sound Vibration 242 (4) (2001) 681699. [2] T. Wang, G. Wang, K. Liu, Simulation control of concrete pump truck boom based on PSO and adaptive robust PD, art. no. 5192810,, Chinese Control and Decision Conference CCDC, 2009, pp. 960963. [3] S. Zhou, S. Zhang, Co-simulation on automatic pouring of truck-mounted concrete boom pump, art. no. 4338699, Proceedings of the IEEE International Conference on Automation and Logistics, ICAL, 2007, pp. 928932. [4] T. Worthington, Trailer pumps, Concrete Construction World Concrete 52 (3) (2007) 6267. [5] J. Kronenberg, Compact four-arm with multiple strengths, Concrete Eng Int 11 (4) (2007) 5859. [6] J. Kronenberg, First delivery of exible, ve-sectioned booms, Concrete Lond 39 (5) (2005) 4647. [7] A.A. Shabana, Flexible multibody dynamics: review of past and recent developments, Multibody System Dynamics 1 (2) (1997) 189222. [8] A.A. Shabana, R. Schwertassek, Equivalence of the oating frame of reference approach and nite element formulations, Int J Non-Linear Mechanics 33 (3) (1998) 417432. [9] C.B. Drab, J.R. Haslinger, R.U. Pfau, G. Offner, Comparison of the classical formulation with the reference conditions formulation for dynamic exible multibody systems, J Comput Nonlinear Dynamics 2 (4) (2007) 337343. [10] D. Kaplan, F.D. Larrard, T. Sedran, Design of concrete pumping circuit, ACI Materials J 102 (2) (2005) 110117. [11] D. Wassink, V. Lenss, J. Levitt, K. Ludema, Physically based modeling of reciprocating lip seal friction, ASME J Tribology 123 (2) (2001) 404412.

Fig. 15. Comparison between the numerical and experimental outputs; the concrete ow (a) and the back pressure (b).

In conclusion, the test rig created plays an important role both in the design of a control strategy and in the verication and analysis phase of the concrete pump boom behavior. In addition, the test rig allows us to observe all the physical phenomena involved in the working of the concrete pump boom.

Fig. 16. Scheme of the data ow between the two numerical models.