modelling nr3 2014

SCIENTIFIC JOURNAL OF THE TECHNICAL UNIVERSITY OF CIVIL ENGINEERING

Mathematical Modelling in Civil Engineering

BUCHAREST

No. 3 - September 2014

Disclaimer With respect to documents available from this journal neither U.T.C.B. nor any of its employees make any warranty, express or implied, or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed. Reference herein to any specific commercial products, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U.T.C.B. The views and opinions of authors expressed herein do not necessarily state or reflect those of U.T.C.B., and shall not be used for advertising or product endorsement purposes

CONTENTS

HIGH STABILITY AND LOW EMISSIONS BURNERS USING KARLOWITZ EFFECT IN CONICAL BURNERS ................................................................................................................................5 Nicolae Antonescu, Paul-Dan Stanescu

INFLUENCE OF DYNAMIC ANALYSIS METHODS ON SEISMIC RESPONSE OF A BUTTRESS DAM ..................................................................................................................................... 16 Cornel Ilinca, Rzvan Vrvorea, Adrian Popovici SIMULATION OF TWO HIGH PRESSURE DISTRIBUTION NETWORK OPERATION IN ONE-NETWORK CONNECTION ......................................................................................................... 31 Sorin Perju, Mdlin Mihailovici, Ioana Stnescu UPDATE OF THE P100-1 CONCRETE PROVISIONS ...................................................................... 40 Viorel Popa

DETERMINATION OF WHEEL-RAIL CONTACT CHARACTERISTICS BY CREATING A SPECIAL PROGRAM FOR CALCULATION ..................................................................................... 52 Ioan Sebean, Yahia Zakaria THERMAL COMFORT ANALYSES IN NATURALLY VENTILATED BUILDINGS .................. 64 Ioana Udrea, Cristiana Croitoru, Ilinca Nastase, Angel Dogeanu, Viorel Badescu

Mathematical Modelling in Civil Engineering, no. 3/2014 5

HIGH STABILITY AND LOW EMISSIONS BURNERS USING KARLOWITZ EFFECT IN CONICAL BURNERS

NICOLAE ANTONESCU Assistant professor, PhD, Technical University of Civil Engineering, Faculty for Building Services Engineering, e-mail: [email protected] PAUL-DAN STANESCU - Professor, PhD, Technical University of Civil Engineering, Faculty for Building Services Engineering, e-mail: [email protected]

Abstract: The conical tunnel burner is an improvement of the cylindrical tunnel burner because, by maintaining all its advantages by means of burning intensification, it may also insure the flame stabilization in a wide range of regulation with considerable diminished pressure losses. Technical applications for the conical furnace burner can vary due to the limited dimensions required by the system, as well as the important thermal loads, and also because of the burning stability characteristic that spreads over an important range of regulation. The low costs required by the burner, generated mainly by its simple construction, also raise the interest for this technical solution. A physical model is proposed for the ignition and flame front stabilization. The flame front stabilization contains different steps from the ignition moment to final flame front stabilization, with specific flame front geometries and specific locations along the burner axis. The installation realised by the authors allowed the experimental study of the burning process in conical furnaces, in order to determine the temperature fields and the flame profiles. The physical model developed for this new type of application and the experimental data sets obtained (along with their interpretation) make the subject of this paper.

Keywords: intensified burning process, stabilization, Karlowitz effect, conical burner

1. Introduction

The cylindrical tunnel burner was the first accomplishment of intensified burning processes and we can mention as that the first historical application of the cylindrical tunnel burner was patented in France in 1907. The patent has been obtained by Traian Vuia, one of the great Romanian scientists, and referred to airplane applications, because of its burning thermal loads up to 1000 times higher than for normal atmospheric burners characterized by laminar planar flame speed.

In our researches we applied the tunnel burner burning process intensification technique to our specific burner geometry. The conical geometry that we chose has revealed important functional and technical advantages, also maintaining the characteristic advantages for the cylindrical tunnel burner intensified burning process.

For the burning processes with lateral confinement it was observed that measured turbulence after the burning process was greater than the determined one from the calculus of flow speeds, even when the molecular and thermal expansion was taken into account. The intensification of the burning process due to the additional turbulization, also called self-turbulization induced by the flame front and known as Karlowitz effect, was theoretically studied for the cylindrical tunnel burners [1]. The process is based on the phenomenon of conversion of flame front transversal speeds energies into turbulent energies (vortexes) because of the flow limitation imposed by the burner walls that doesnt allow lateral expansion. This kind of process is generally called with non-slip boundary conditions at the wall and from different researches [2] [5] , it resulted that the burning process speed rises up to 10 times due to self turbulisation in cylindrical tunnel burners by respect to laminar planar flame speed. In the study Turbulent burning, flame acceleration, explosion triggering [6] the author indicates that only by taking into account the effect of non-slip boundary conditions in addition to general flow turbulence, the model for the burning process (applied to cylindrical burners) can match the experimental results.

6 Mathematical Modelling in Civil Engineering, no. 3/2014

2. Construction and functioning principle

The conical tunnel burner is fed with air-combustible mixture in the kinetic dosage domain through a tube with interior diameter smaller than the minimal ignition diameter, in order to avoid the flame return phenomenon. In the case of high speed flows of combustible mixtures the minimal ignition diameter is considerably larger then for the case of stationary ignition, fact resulting from the ignition equations analyze. The construction and functioning principle of the conical burner is described in Figure1.

The gaseous fuel and the necessary air for the burning process are fed in the upstream end of a homogenization element realized from a cylinder filled with non-aerodynamic geometry bodies randomly disposed. The mixture must be in the kinetic burning domain and well homogenized, in order to reduce flame stabilization variations.

1. conical refractory body; 2. homogenization element; 3. gaseous fuel intake;

4. air intake; 5. photo of the working burner (opened end)

Fig 1 - Conical tunnel furnace

At the moment of superior ignition, at the opened end of the conical burner, the flame front will move upstream (towards the intake nozzle of the burner) in a stable position, characteristic for the cold walls ignition process. The stability condition is reached if the turbulent burning speed in the wall vicinity (considering the extinguishing effect generated by the cold surface) is equal to the flow speed in the hydraulic boundary layer (considering speed variation from zero to mean stream value).

In the very next moment the flame front additional turbulence appears due to non-slip boundary conditions at the wall imposed by the conical geometry of the burner. This increase of the burning speed generates a further upstream movement of the flame front in a new stable ignition position where, due to the conical form of the burner, the flame turbulization further grows, and so on, till the flow speed equalize the enhanced burning speed due to the Karlowitz effect. The equilibrium between the burning process intensification and marginal detachment of the flame due to upstream flow speed increase generated by the conical form of the burner is, generally

5

combustion air

1

2

3

4

gaseous fuel


speaking, a stable process, fact experimentally demonstrated by the real stabilization of the flame front at a certain axial distance from the intake nozzle of the conical burner.

As the wall temperature of the burner increases, due to heat transfer from the flue gases and the flame front, at a certain moment the combustible mixture ignition temperature will be reached and the flame front will adhere to the solid surface getting a new stable ignition position. The phenomenon between the primary ignition and final stabilization happens in a very short delay, by the order of seconds, due to high speeds and intense heat transfers involved. For the moment, in this experimental phase, those transient states will not be detailed. What was experimentally determined was that a few seconds after primary ignition the burning process is stable and fixed at a certain axial distance from the intake nozzle of the burner. The flame front stabilization is self regulated on different sections, closer to the intake nozzle, on smaller diameters of the cone, when debits diminish, and further from the nozzle when debits rise.

3. Physical model

A physical model can be deducted for describing the ignition and flame front stabilization. The flame front stabilization contains three different steps from the ignition moment to final flame front stabilization, with specific flame front geometries and specific locations along the burner axis. The evolution of the process is described in Figure 2.

In the first stage of the process (see Figure 2a), the burner is cold and the gaseous combustible mixture is ignited at the exhaust end of the burner by an external source (spark or hot gases). The flame front moves upstream towards the intake nozzle because flame return conditions are present, meaning that at least for one geometrical contour the speed of the turbulent burning process exceeds the speed of the flow (due to boundary layer speed distribution effect).

a. first stage of the process b. stabilization by the equality c. equal speed stabilization between flow speed in the boundary turbulent speed completed layer and the increased by hot surfaces ignition

Fig 2. - The three different types of flame front between ignition and final stabilization


The gradient of the turbulent burning speed in the boundary layer zone has two different domains:

near the wall, on a certain distance, the extinguishing effect of the cold wall is present and so the burning process is blocked because the flammable mixture temperature is reduced under the ignition point; for these zone the burning speed is null;

beyond the cold wall extinguishing effect zone, the turbulent burning speed rapidly varies from zero to nominal values, due to the exponential dependence of the burning speed with the combustible mixture temperature;

The first stabilization contour may be point out for the equality of turbulent burning speed with the flow speed in the boundary layer.

In the second stage (see Figure 2b) the walls of the burner are still cold and the flame front stabilization corresponds to the first step model. Because of the lateral flow limitation, the Karlowitz effect appears, generating the self-turbulization of the flame front. The result is an increase of the turbulent burning speed due to the additional turbulence and so the flame front further migrates towards the intake nozzle, on a smaller section zone of the conical burner. The flame front is stable and the stabilization contour conditions are now generated by the equality between flow speed in the boundary layer and the increased turbulent speed, characteristic for the burning process after getting the additional Karlowitz effect turbulence. In the third stage of the flame front stabilization, the burner wall reaches the ignition temperature for the gaseous combustible mixture, more precisely for the studied case the 780 oC value, and the flame front gets stabilization conditions directly on the burner wall. In this case, the equal speed stabilization conditions are completed by the hot surfaces ignition conditions. In the picture representing the third stage of flame front stabilization (see Figure 2c) it may be easily observed the plane form of the flame front that respects the speed distribution of the flow in the section and the upwards oriented stabilization contour due to direct ignition from the wall. The hot wall surface ignition comes in addition to turbulent stabilisation. The additional turbulization due to Karlowitz effect is well present in a conical geometry burner, although a certain lateral expansion is permitted to the flue gases in the flame front (the non-slip boundary condition is not total, as in the case of cylindrical burners). This is happening because this lateral expansion allowed by the burners conical geometry is significantly less important than the expansion tendency generated by the dilatation of the air-combustible mixture in the burning process.

So, for the conical burner solution, the lateral constriction of the flow is present as in the case of the cylindrical tube burner, maintaining the intensified character of the burning process, but also bringing some important specific advantages:

due to the conical geometry of the burner that generates a continuously wider flow section, a self stabilization of the flame front in a certain section along the flow is possible, unlike the case of the cylindrical burner where special stabilization systems must be used, otherwise the flame front will reach the combustible mixture intake nozzle;

the self-stabilization process is valid in a wide regulation domain for the combustible mixture flow;

the continuously wider flow section of the burning chamber leads to flow deceleration along the axis of the burner by comparison with the cylindrical burner and so the hydraulic pressure losses are significantly lower, with no significant implication for the stability process, because the self-turbulization is a local process and the widening of the flow section is negligible along the flame front depth;

the flame front stabilization at the cylindrical burner may be realized only by the means of special geometry elements like peripheral recirculation chambers or diaphragms,


unlike the conical burner that self-stabilizes the flame front, due to the variation of the flow section, at a certain distance from the intake nozzle, where the intensified turbulent burning speed of the mixture equals the mean speed of the flow;

due to the fact that the combustible mixture enters through the entire section at the cylindrical burner, its stability limit is linked to the back-flame situation, unlike the case of the conical burner where the small intake section of the nozzle is characterized by flow speeds bigger than the turbulent burning speed even at low thermal charges.

As for the debit variation situation, due to the conical geometry of the burner, the flame front stabilization is self regulated on different conical tube diameters and so, if the combustible mixture debit decreases, its flow speed will also decrease and the stabilization diameter will also decrease, corresponding to a closer to the nozzle position than the previous one. Oppositely, if the debit rises, the stabilization process will move downstream along the burner axis.

The angle of the burner cone is limited to 12o, the detachment angle for free axial streams. If the angle surpasses this limit value, a Craya-Courtet boundary recirculation forms, generating a recirculation stabilization effect instead of a self-turbulization Karlowitz effect stabilization. Along with the increase of the cone angle the stabilization process will get closer to the case of wide furnaces burners stabilization, with lower burning speeds and lower furnace thermal loads.

4. Equational model

At the ignition moment the flame front moves upstream in the conical tube due to back-flame conditions that appears because the free flame turbulent burning speed ut is greater than the flow speed w anywhere in the exit section. The variation curves for the two speed values are represented in Figure 3.

The gradient of the flow speed in the boundary layer may be accepted as linear (Prandtl and Taylor) , described by the equation [1] :

2.0

8.08.1

023.0D

wg mT

(1) where: gT is the speed gradient in the boundary layer, wm is the mean value of the flow speed, is the cinematic viscosity and D is the tube diameter. The gradient of the turbulent burning speed in the boundary layer zone has two different domains:

near the wall the extinguishing effect is present and so ut = 0 ; beyond the cold wall extinguishing effect zone the turbulent burning speed rapidly varies

from zero to nominal values, due to the exponential dependence of the burning speed by the combustible mixture temperature [7] (Bollinger):

4.1

2

1

2

1

TT

uu

T

T

(2) Due to wall limited flame conditions the Karlowitz effect of burning speed raise takes place. Compared to the process in cylindrical geometry tunnel burners, the process in conical tunnel burners is affected by the flow section relaxation on radial direction, due to the specific geometry. The phenomenon for the conical geometry is illustrated in Figure 4.


Fig 3. - The variation curves for the flow speed and for the burning speed

On a perpendicular direction on the momentary flame front, flow acceleration appears, due to burning process with flue gas formation and temperature raise from ignition temperature to theoretical burning temperature. In the hypothesis of a burning process with no chemical relaxation, the flame front acceleration from normal combustible mixture flow speed u0 (and density 0 ) to flame front flue gases flow speed uT (and density T ) , is given by the speed increase defined as:

T

Tfr uw

00(3)

This acceleration may effectively develop on axial direction (considering the general flow) but on radial direction it is confined by the solid walls of the burner and may not develop. If the walls are of cylindrical shape, the confinement is complete, but in the case of a conical geometry, a certain lateral expansion is permitted, in the limits of section increase due to /2 angle made by the wall with the main flow axis, as shown in Figure 4.

Fig 4. - Flame front in conical tube burner

The axial acceleration generated by the axial speed difference wax has the expression:

cos00

T

Tax uw

(4) where is the flame front mean angle, as defined in the Damkhler turbulent burning model [7]. Considering the lateral expansion allowed by the conical geometry, a radial acceleration can be considered as given by the speed difference on radial direction wr :

/2 wr conical shape allowed radial component

wr flame front movement radial component

uT

wa

/2

u0

x x x

w

UT = w

UT UT w


2

tancos00

T

Tr uw

(5) Because in the Damkhler turbulent burning model the angle is defined by the relation cos = uo/uT the speed differences determining the axial and radial accelerations becomes:

TT

Tax u

uuw 000

(6) and

2

tan000

TT

Tr u

uuw (7)

The total energy generated by the flame front acceleration is proportional with the square power of the speed difference:

2002

T

Tfrfr uwE

(8)

From the total acceleration energy generated by the flame front, only a part is consumed by the means of the axial component of the speed difference and by the means of the allowed (by the conical shape) radial speed difference energy and converted into effective gain of kinetic energy for the flow. The remaining energy, that transforms into additional turbulence (Karlowitz effect for conical geometries) will be the difference between total acceleration energy generated by the flame front and total kinetic energies gained by the flow both on axial and radial directions:

2tan1

2tan

22

02

02

00

200

0

200

0

20

0

TTT

TT

TT

T

TT

T

T

TT

uu

uu

uE

uu

uuu

uuE

(9) and therefore:

2tan1

31 2

20

200

0TTT

T

T

TKuu

uuu

uu

(10) where: uTK is the intensified turbulent burning speed, uT is the free turbulent burning speed and is the turbulence intensity of the room temperature combustible mixture flow.

5. Experimental works

The range of flows for witch the flame stability is insured by the conical burner, especially when the combustible mixture debit diminishes, may be considered very wide. It is important to highlight the characteristic of the conical burner of not generating back-ignition by comparison with the cylindrical tunnel burner where the flame return generated by the appearance of Karlowitz effect can be stopped only if special bodies are inserted in the burner, bodies characterized by a flow sections smaller than the minimal ignition diameter.

The experimental installation had the following main characteristics: gaseous fuel intake pipe interior diameter : 6 mm; angle of the conical body of the burner : 12o;


refractory body made from cement with ceramic fibers; combustible : natural gas G20; mixture dosage : stoychiometric; gaseous combustible debit : 200 l/h and 100 l/h; burner heat output: 2000 W respectively 1000 W.

In figure 5 is presented a photo made from the opened end (a) and the corresponding thermography (b).

a b

Fig 5. - Flame photo made from the opened end (a) and the corresponding thermography (b).

First of all it must be stated that the 2000 W thermal load for the burner did not generate any flame front braking or suspension, the burning process being stable during the whole experiment. Extrapolating the dataset we may appreciate that a conical burner with an intake nozzle of only 42mm will reach a thermal load of 100 kW.

The experimental research aimed to establish the temperature fields for the conical burner. Those fields provide basic information regarding the flame front stability, the burning speed and the thermal load for the burners flow section. The measures were performed with a microthermoelement placed at different levels from the intake nozzle at different radial distances. The axial levels were between 5 and 40 mm with a 5 mm distance step and the radial positions were distributed from the centre of the section (axial position) to the conical wall with a 2 mm step. The measured temperature values for a semi section of the burner, generated by the flame front configuration and by the burning process zone, are presented in Figure 6.

The important intensification of the burning process is demonstrated by the short distance after intake that characterizes the ignition zone, situated at 15 mm axial distance from the nozzle. The wall extinguishing effect is visible and at a radial distance from the wall of only 2 mm the temperatures reaches the ignition level of approx. 800 oC. The radial temperature field for this section is quite constant around this temperature, meaning that the whole section is in fact an ignition section. This geometry corresponds both to speed distribution in turbulent flows and also to the expected ignition phenomenon characterized by Karlowitz effect of self-turbulization. The burning process develops, fact demonstrated by the raise of the temperatures, and reaches a maximum values distribution curve at an axial distance of 25 mm. The 10 mm distance between ignition section and burning process ending may be assimilated with the flame front thickness. The 10 mm value for flame thickness was also expected for this type of intensified burning process, confirming once more the presence of the Karlowitz effect.


The exponential trend of the temperature from ignition to maximal value, experimentally demonstrated by the temperatures measured in the 20 mm distance section, also confirms the burning process development, known as having these evolution from other different turbulent flame studies [8].

Fig 6 - Burner geometry (photos) , distribution of measuring points and temperature fields

By expressing the experimental datasets in axial evolution for different radial distances from the axial position it may be also visualized the ignition section and the end of burning process section. In Table 1, the temperature level of 700 800 oC, corresponding to the ignition point for the gaseous combustible mixture, is reached at an axial distance of 15 mm, for axial distances between 2 and 8 mm. The maximum values for the temperatures are reached at an axial distance of 25 mm for all radial coordinates. The bold marked zone in the table corresponds to the burning process.

The thickness of the flame front in a turbulent burning process grows along with the turbulent intensity as a result of the flame front local curving (Damkhler model) and due to turbulent diffusion that displaces burning volumes that may constitute in local ignition islands (Growe model). The additional turbulence generated by the Karlowitz effect in the conical tube is the explanation for the 10 mm flame front thickness value, corresponding to high levels of turbulence intensity, instead of the 2 3 mm flame front thickness that represents the normal values for the free flames at the same flow speeds.

Table 1

Axial downstream temperatures evolutions at different radial coordinates

H=5 H=10 H=15 H=20 H=25 H=30 H=35 H=40

R=2 20 90 830 1160 1340 1180 1050 920

R=4 20 70 780 1400 1440 1380 1260 1220

R=6 20 90 860 1430 1540 1500 1440 1390

R=8 20 70 690 1450 1580 1550 1490 1460

After the 25 mm section, the measured temperatures decrease, due to the lack of internal heat sources and because of the burner heat losses presence.

Gas debit = 200 l/h ; air excess number = 1

0

200

400

600

800

1000

1200

1400

1600

1800

R=0 R=2 R=4 R=6 R=8 R=10 R=12radial distance from the burner wall [mm]

flue

gas

tem

pera

ture

[C]

H=5H=10H=15H=20H=25H=30H=35H=40


The flame front geometry (see Figure 2) is clearly plane in the section outside the boundary layer of the burner wall. The explanation for these specific flame front geometry is linked to the conical geometry of the burner along with the high intensity turbulence of the process that determines the burning process to develop on the uniform speed distribution existing in the flow, unlike the case of free flames were the flame shape is conical with the edge on the flow axis.

The plane form of the flame front, generated by the equality of the burning speed with the flow speed, states for the high thermal loads that characterize the section of the burner. For the experimental burner, on the flame front stabilization section of 1,2 cm2, the medium flow speed generated by the 2200 l/h of gaseous fuel mixture is about 5,09 m/s. This speed, characteristic for the conical geometry, self intensified, turbulent burning process, compared with the planar flame speed for methane stoychiometric mixtures that is 0,36 m/s, indicates a process intensification of about 14 times. This value is normal [6] and characteristic for the tunnel burners with Karlowitz effect and proves that the advantages induced by the conical form of the burner does not affect on its efficiency by respect to the burning process intensification.

As shown in figure 7 by the temperature fields for the case of 100 l/h gaseous fuel intake, the general features of the process does not change. So, by realizing the measures with the same geometrical pattern as in the case of 200 l/h gaseous fuel intake, the same isothermal curves are obtained, stating that the burning process intensification by means of Karlowitz effect is also present at lower thermal loads.

Fig 7 - Temperatures fields measured values for two thermal loads

6. Conclusions and discussions

From the point of view of supplementary turbulization generated by the non-slip boundary conditions at the wall the flow section enlargement generated by the conical geometry of the burner can be neglected in terms of overall axial deceleration. Anyhow, the component of speed difference projection on radial direction (see figure 4 and equation 5) will induce an increase of the flow energy, but compared with the axial flow energy, the increase is quite of little importance, due to the cone angle that is limited to 12o.

Calculating:

2

tan 2 0.011 for

2 6o, it results an increase of the axial flow energy of

maximum 1,1 % for the conical burner case, compared with the cylindrical burner case. So, the turbulent burning speed increase, which is directly linked to the self-turbulization process, will reach the same values in either cylindrical or conical burner.

BURNING PROCESS : Dair = 2000 l/h, Dg = 200 l/h

0

400

800

1200

1600

0 2 4 6 8 10 12 14 16 18 20 22 24radial distance [mm]

[C]

H=5H=10H=15H=20H=25H=30H=35H=40

BURNING PROCESS : Dair = 1000 l/h ; Dg = 100 l/h

0

400

800

1200

1600

0 2 4 6 8 10 12 14 16 18 20 22 24radial distance [mm]

[C]

H=5H=10H=15H=20H=25H=30H=35H=40


The conical tunnel burner is an improvement of the cylindrical tunnel burner because, by maintaining all its advantages by means of burning intensification, it may also insure the flame stabilization in a wide range of regulation and with considerable diminished pressure losses.

We consider that the first applications of the conical tunnel burner will be for boilers, especially for condensing boilers, but also for furnaces and other technological applications that require large regulation domains and quick thermal responses and maybe even for gas turbines burning chambers.

The important self-turbulization that occurs in the conical burner makes the burning process speed to increase, for methane-air stoichiometric mixture, from the usual values of 1 to 2 m/s, characteristic for turbulent free flames, to 5 m/s. In those conditions the sectional thermal load for the conical burner reaches values averaging 18 MW/m2.

The ignition of the conical furnace burner can be realized in very stable conditions because of its geometry with variable flow section that can determine the existence of a stable ignition contour in the boundary layer of a flow section between the infeed nozzle and the exit section. This fact is particular for the conical burner, the cylindrical burner having no possibility of stabilizing the flame front somewhere along the flow without proper special stabilization devices such as annular recirculation chambers or anti-backflame structures.

The flame front in the conical furnace burner is plane and very stable, the backflame phenomenon being (in general, for the tested range of thermal powers) constructively avoided, without the need of any additional devices.

Because of the stabilization process, priory described, the conical furnace burner allows wide regulation domains (down to 1:10 of the nominal load), significantly wider than in most technical cases (1:2 of the nominal load for normal forced draught burners and 1:5 of the nominal load for vortex burners). Practically, the regulation domain for the combustible mixture debits is given by the ratio between the surface of the exit section of the burner and the surface of the intake section of the burner, just after the infeed nozzle.

An important advantage is the low pressure losses that characterize the conical burner compared to the tunnel burner (cylindrical). The fact is due to the flow section enlargement after stabilization of the burning process that allows the deceleration of the flue gases debit.

In further researches we intend to enlarge our study of conical furnace burners applications for the cases of other thermal loads and types of boilers, for direct mixture air heaters and also for energetical facilities.

This paper was presented at EENVIRO 2014 Conference.

References

[1] Lewis B., Pease R., Taylor H. (1956). Combustion processes, Ed. Princeton Univ. [2] Akkerman V., Bychkov V., et al. (2006). Flame oscillations in tubes with non-slip at the walls, Combustion and

Flame 145, (pp. 675 687). [3] Bychkov V., Petchenko A., Akkerman V., et al. (2005). Theory and modeling of accelerating flames in tubes,

Phys. Rev. E 72. [4] Akkerman V., Bychkov V., Petchenko A., et al., (2006). Accelerating flames in cylindrical tubes with non-slip at

the walls, Combustion and Flame 145, (pp. 206 219). [5] Bychkov V., Akkerman V., et al., Flame acceleration at the early stages of burning in tubes, Combustion and

Flame. [6] Akkerman V. (2007). Turbulent burning, flame acceleration, explosion triggering Umea University

Department of Physics. [7] Antonescu N., Stanescu P.D., Antonescu N.N. (2002). Burning processes - theoretical and experimental bases,

publishing house Matrix-Rom Bucharest, 317 pages. [8] Antonescu N., Burning processes with peripheral stabilization in tunnel furnace burners - Energetical Studies and

Researches publishing house of the Romanian Academy - tome XII, nr.4, (pp. 416-434).


INFLUENCE OF DYNAMIC ANALYSIS METHODS ON SEISMIC RESPONSE OF A BUTTRESS DAM

CORNEL ILINCA - Lecturer, PhD, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected] RZVAN VRVOREA PhD student, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected] ADRIAN POPOVICI - Professor, PhD, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected]

Abstract: The seismic analysis of a buttress dam with 73.50 m height is performed by the spectral analysis method and the direct time integration method. An accelerogram with 0.1g maximum acceleration was applied horizontally, in the upstream - downstream direction, at the bottom of the dam-foundation finite element mesh. The hydrodynamic effect of the reservoir was considered according to the added mass procedure (Westergaard relation). ABAQUS software was used to make the analyses. The same type of finite element C3D20R was used for the mesh of the dam body and of the foundation. The comparison of the results is made on the displacements, the stress state and the sliding stability on the dam-foundation contact in the full reservoir hypothesis. The comprehensive analysis concluded that both methods had provided close results for the considered case study. The spectral analysis method revealed itself to be more conservative compared to the direct time integration method.

Keywords: dams, earthquakes, dynamic analyses, stress

1. Introduction

Edward Wilson a well-known Professor of seismic analysis of structures from the University of California wrote (August, 28th 2013, www.edwilson.org) the following comments on seismic analysis methods: Convince engineers that the Response Spectrum Method produces very poor results. Convince engineers that it is easy to conduct Linear Dynamic Response Analysis. The scope of this paper is to check this opinion in the case of concrete dams, structures known as having a complex seismic response because of the interaction of the structure with water from the reservoir and the foundation area [6], [7], [8]. The seismic response of the Gura Rului dam, a buttress dam with 73.50 m maximum height, is evaluated comparatively by means of the Response Spectrum Method and the Direct Time Integration of Motion Equations [1], [4]. The seismic action with a PGA of 0.1g consisted of a horizontal component of the accelerogram recorded at Focsani seismic station during the August 30th 1986 Vrancea earthquake. The horizontal component with a PGA of 0.271g (g-gravity) of the recorded accelerogram at Focsani INCERC seismic station during the August 30th 1986 Vrancea earthquake, scaled to 0.1g maximum acceleration was used to act on Gura Rului dam. The subsoil of the seismic station has a geologic profile close to that of the dam site. The accelerogram was applied at the bottom of the damfoundation system finite element mesh horizontally, on the upstream-downstream direction. The influence of the reservoir on the seismic response was taken into account by the added mass procedure. The added masses were computed according to the Westergaard relation. The analysis was performed with ABAQUS software, the solid elements type C3D20R were used for the mesh of the damfoundation system. The linear elastic behavior of the materials was accepted in all analyses [3]. The comparison of results in the dam seismic analysis for those two methods mentioned above is performed in displacements, stress state and sliding stability on the dam-foundation contact in the full reservoir hypothesis.


2. Short description of the Gura Rului dam

Gura Rului dam, commissioned in 1974, is a buttress dam with 73.50m maximum height (Fig.1). [2], [5].

Fig. 1 - Gura Rului dam layout (a) and longitudinal section through the dam axis (b): 1 non-overflow block, 2 overflow block, 3 buttress, 4 crest, 5 overflow span, 6 stilling basin

The dam is located within the Sub-Carpathian Depression of Sibiu County. The rock from the dam foundation consists generally by gneiss with rare amphibolite intercalations and pegmatite intrusions. The friction coefficient to sliding on the dam-foundation contact was evaluated at 0.70. Gura Rului development has multiple uses as follows: water supply of Sibiu city and neighboring localities (1440 l/s), production of hydroelectric energy (hydropower plant with Pi=3700 kW located downstream of the dam) and flood control [2].

The main characteristics of the development can be summarized as follows (Fig. 2):

- maximum height of the dam.. 73.50 m - crest length... 328.00 m - crest width.... 6.20 m - base width.. 57.40 m - upstream and downstream slopes.1 : 0.57 and 1 : 0.28 - thickness of buttress (variable)...4.508.00 m - concrete volume of the dam body.. .300500 m3 - capacity of discharge works.832 m3/s - reservoir volume at NRL15106 m3

The dam blocks are with polygonal heads and buttresses with variable thickness in horizontal cross-section, increasing from the contact polygonal head-buttress to downstream. This innovative solution proposed by Priscu Popovici [1], applied for the first time in this field led to significant savings in dam concrete volume evaluated at about 14% compared to the usual solution with buttresses having constant thickness.


Fig. 2 - Representative cross sections through block no. 11

A number of 19 non-overflow blocks and 3 overflow blocks make up the retention. Water tightening between the blocks within the area of the polygonal head units was achieved by means of a 1.5 mm thick copper steel sheet doubled by a M35 type polyvinyl band.

The dam spillway consists of three free overflow openings located in central blocks. The total length of the surface spillway is 39 m with maximum discharge capacity of 800 m3/s. The bottom outlets consist of two pipes with 1000 mm diameter having together a total discharge capacity of 32 m3/s with the reservoir at Normal Retention Level (NRL). The energy of the discharged water from the reservoir is dissipated in a battery consisting of two stilling basins located at the dam downstream heel and having 34+18=52 m total length. The foundation sealing was carried out by a grout curtain consisting of two rows of injected drillings at 1.50 m span, with maximum depth of 45 m for the upstream row and 40 m for the downstream row. In order to discharge the uplift pressure acting into the dam foundation, downstream of the grout curtain a row of 30 m deep draining drillings displayed by two for each block was provided. According to the Romanian national regulations, the dam is classified as the second class in terms of the economic importance and in the category B in terms of the collapsing risk. Thus, this dam requires a special monitoring during its lifetime. In order to achieve this provision, the dam-foundation unitary system was equipped with several monitoring devices (4 direct pendulums, 2 inverted pendulums, 7 rock meters with 3 rods, piezometer drillings, hydrometers, telepress meters etc.). The monitoring activity, accompanied by visual inspections, ensures the proper surveillance of the dam behavior. It should be mentioned that the dam has behaved normally in terms of displacements, stresses, and seepage during the entire operation, which started in 1974.

3. Mathematical models and input data

The block no. 11 (Fig. 2) was selected to be computed to OBE seismic action.

In compliance with the Seismic Hazard Map of Romania the dam site is located in an area with PGA = 0.20 g (PGA Peak Ground Acceleration) and period corner of 0.7 s.

Table 21 from NP 0762013 sets that for dams classified in the II class as importance and B category as collapsing risk OBE (Operation Basis Earthquake) is 0.28 PGA but no less than 0.10g. Consequently, the dam was computed to the action of an earthquake with 0.10g maximum acceleration.


Fig. 3 - N-S component of recorded accelerogram at Focsani INCERC station during the 30th August 1986

Vrancea Earthquake

The seismic action consisting of NS component of recorded accelerogram at Focsani INCERC station during the 30th August 1986 Vrancea Earthquake (Fig. 3) was applied at the bottom of the damfoundation finite element mesh system, horizontally, upstream-downstream direction.

Fig. 4 - Finite element mesh of dam foundation system: general (a) and dam detail (b)


The finite element mesh is illustrated in Figure 4. Quadratic hexagonal finite elements type C3D20R from ABAQUS software library were used to perform mesh. In Table 1 are given some data on the finite element mesh.

Table 1

Data on finite element mesh

Nodes Elements Subsystem

Total number of nodes

Total number of finite elements

Dam 67118 12835 Foundation 73742 15260

C3D20R is a three-dimensional solid element with 20 nodes and 3 degrees of freedom in each node. It may be degenerated to an element with 9 nodes. This element is a general-purpose quadratic brick element, with reduced integration points (2 x 2 x 2 integration points). The reduced integration uses a lowerorder integration to form the element stiffness but reduces running time, especially in three dimensions. The mass matrix and the distributed loadings use full integration.

The seismic response spectrum was computed from the accelerogram given in Figure 3, using a well-known relation with convolution integral:

|| 1

sin

1

Respectively Sd (, ) = , , (2)

Where ||is the maximum value (spectral) of the seismic response in displacements; circular eigenfrequency of the oscillator ; () accelerogram of the earthquake; - oscillator fraction of the critical damping; Sd, Sv, Sa spectral values of the response in relative displacements, relative velocities and, absolute accelerations respectively.

Fig. 5 - Design spectra computed from the NS component of the accelerogram presented in figure 3

Design spectra were obtained by smoothing the values from the seismic response spectra according to the rule of the least squares. The design spectra for equal to 0.00, 0.02, 0.05, and 0.10 are presented in Figure 5. The seismic analysis of the dam was performed for = 0.05. This value is recommended in literature, resulting from several in dam site dynamic investigations.


The material characteristics of the dam - foundation system are given in Table 2. The materials were considered homogeneous, isotropic and linear elastic behavior.

Table 2

Material characteristics of the dam foundation system

Materials Properties

Concrete Rock foundation

Mass density (kg/m3) 2400 - Static Poisson coefficient 0.16 0.26

Dynamic Poisson coefficient 0.22 - Static Young modulus (MPa) 24000 20000

Dynamic Young modulus (MPa) 32400 -

The effect of water in the seismic response was considered by the added mass procedure. The definition of the added mass [Mh] acting on the normal direction in the point of application is as follows:

{Ph(t)} = - [Mh] {+ r}n (3) where {Ph(t)} is the hydrodynamic force in the reference point and {+r}n is the total acceleration response to the normal direction at surface in the reference point.

This means that added mass as value corresponds with the hydrodynamic force generated by a unitary acceleration on the normal direction at surface in the reference point.

Fig. 6 - Assessment of added masses

In the general case when the directions of the normal to surface, of the earthquake and of the structure degrees of freedom are different, the added mass computed according to relation (3) must be projected successively on earthquake direction and on the structure degrees of freedom (Fig. 6). The dynamic equilibrium equations are written as follows:

[[M] + [Mh] [rcn] [rn,x,y,z]] { r} + [C] { r} + [K] { r} = -[[M] + [Mh] [rcn] [rn,x,y,z]]{r} (4) where [rcn] has the dimensions equal with the number of the degrees of freedom of the system and contains on the diagonal the cosine directors between the normal to surface in the nodes of the mesh and earthquake direction;

[rn,x,y,z] has dimensions corresponding to the number of the degrees of freedom of the system and contains on the diagonal the cosine directors between the normal to surface in the nodes of the mesh and directions of the system degrees of freedom.

The boundary conditions in the static and spectral analyses consisted in the displacement blockage in x, y, z directions for all nodes located on the lateral faces and the bottom of the foundation and in the y direction of the bank, bank lateral faces of the profile polygonal head.

In the direct time integration the accelerogram was applied at the bottom of the mesh on x direction. The displacements on y, z directions at the bottom of the mesh, y directions on the bank, bank lateral faces of the mesh (dam plate and foundation) and x, y, z directions on upstream, downstream lateral faces of the foundation respectively were blocked.


The time step (t) in direct time integration was constant and equal to 0.01 s. The response accelerations during a time step were computed using an implicit scheme by solving for dynamic response parameters at time t + t based not only on values at t but also on these same quantities at t + t. But because they are implicit, the nonlinear equations must be solved. The static loads and the seismic action were applied successively in distinct steps: dead load (step 1), hydrostatic pressure (step 2), uplift pressure under polygonal head (step 3) and seismic action (step 4).

In the direct time integration analysis, the damping matrix [C] was evaluated as a function of the mass matrix [M] and stiffness matrix [K] in compliance with the linear Rayleigh relation:

[C] = [M] + [K] (5) where and coefficients were evaluated for = 5% in first two mode shapes in full reservoir hypothesis ( = 1.402 and = 0.00158). 4. Results in free vibration analysis

In Table 3 the first six circular eigenfrequencies (, rad/s) eigenfrequencies (f, Hz) and eigenperiods (T, s) of the dam profile are presented in both empty and full reservoir hypotheses.

Table 3

Free vibration characteristics

Mode number

Empty reservoir Full reservoir (NRL level) Eigenfrequency Period Eigenfrequency Period

(rad/s) f (Hz) T (s) (rad/s) f (Hz) T (s) 1 30.4 4.834 0.207 21.0 3.343 0.299 2 47.0 7.492 0.133 42.4 6.751 0.148 3 59.9 9.533 0.105 45.7 7.275 0.137 4 64.6 10.280 0.097 46.3 7.375 0.136 5 91.7 14.595 0.069 71.6 11.397 0.088 6 107.2 17.069 0.059 85.2 13.566 0.074

In Figure 7 are illustrated first six lowest mode shapes in full reservoir hypothesis.

T1=0.299 s T2=0.148 sT3=0.137 s

T4=0.136 s T5=0.088 sT6=0.074 s

Fig. 7 - Gura Rului dam first six lowest mode shapes in full reservoir hypothesis (reservoir level at NRL)


It should be noted that Gura Rului buttress dam is classified in the field of rigid structures. The fundamental dam mode is developed mainly on the horizontal upstream downstream direction (x axis). The second mode in the empty reservoir hypothesis and the fourth mode in the full reservoir hypothesis are developed horizontally along the dam crest (bank bank) modeling the free vibration of the dam buttress as a plate (y direction). The reservoir effect led to the increase of the dam fundamental period with 44% compared to the empty reservoir (0.299 s versus 0.207 s), but the profile remains in the same domain of rigid structures.

The participation factors in six lowest mode shapes are given in Table 4. The participation factor for a mode k in direction i is a variable which indicates how strong the global x,y,z translation is about one of the three axes represented in the eigenvector of that mode. For instance, the fundamental mode is important on x axis (horizontal upstream downstream direction)

Table 4

Eigenmodes and participation factor (empty and full reservoir)

Empty reservoir Full reservoir

Mod

e N

o.

Natural frequency Eigenperiods Participation factor

Mod

e N

o.

Natural frequency Eigenperiods Participation factor

[Hz] [s] x y z [Hz] [s] x y z

1 4.834 0.207 2.15 0 -0.20 1 3.343 0.299 -7516 0 -719

2 7.492 0.133 0 2.05 0 2 6.751 0.148 1596 0 -8804

3 9.533 0.105 0.43 0 1.46 3 7.275 0.137 -4467 8 -2273

4 10.28 0.097 -1.67 0 1.39 4 7.375 0.136 -11 -3142 -7

5 14.60 0.069 0 -0.43 0 5 11.40 0.088 -2403 0 812

6 17.07 0.059 0 -0.93 0 6 13.57 0.074 1 -1040 1

5. Analysis Results by response spectrum method

Response spectrum analysis provides an inexpensive approach to estimating the peak response of a structure subjected to base motion: the simultaneous motion of all nodes fixed with boundary conditions.

The maximum dam displacements and stresses ||max were computed from m (m=25) representative peak modal responses using the square root of the sum of the squares (SRSS):

||max =( || 2)1/2(6) where index i is referring to the degree of freedom and k to the eigenmode.

zxxz Fig. 8 - z, x and xz contour lines in the central section of the profile resulted in spectral analysis,

full reservoir (kPa)


In Figure 8 z, x and xz contour lines are illustrated in kPa in the central section of the profile resulted in the spectral analysis, full reservoir hypothesis. The maximum vertical stress (z) reaches maximum value of 1277 kPa at downstream face. The maximum shear stress reaches maximum value of 489 kPa at downstream dam heel. The alternative maximum displacement on upstreamdownstream horizontal direction computed by the spectral analysis reaches 9 mm at crest. On the vertical direction it reaches 2 mm.

The variations in elevation of z and xz resulted in the spectral analysis, full reservoir case, at upstream and downstream faces of the profile are illustrated in Figure 9.

Fig. 9 - Variation of z and xz along the upstream and downstream faces of the dam profile, full reservoir case: GP

dead weight,. Sdead weight + hydrostatic pressure including uplift, Cspectral analysis 0.1 g

It can be remarked that vertical stresses z due to load combination dead weight + hydrostatic pressure including uplift spectral stresses 0.1 g vary between (-1000+600) kPa at upstream face and (-3900+300) kPa at downstream face (+ tensile stress).

Shear stresses xz directed from downstream to upstream in the dam body at the same load combination reach 500 kPa at upstream face and 1100 kPa at downstream face.


Fig. 10 - z and xz diagrams on dam foundation contact computed in spectral analysis in different hypotheses in

full reservoir case (S dead weight + hydrostatic pressure + uplift pressure, C spectral analysis, ME elementary method computing stresses + inertia forces computed in compliance with spectral analysis)

Some representative diagram of z and xz at different hypothesis on dam foundation contact in full reservoir case are given in Figure 10. The meaning of notations S, C, was already given, ME corresponds to elementary method computing stresses + inertia forces computed in compliance with spectral analysis. A satisfactory correlation should be noted between z diagram at S-C combination obtained in the spectral analysis and the one obtained by the elementary method.

Fig. 11 - Scheme of loads taken into account in the analysis based on strength materials theory (elementary method

computing stresses + inertia forces computed in compliance with spectral analysis)

4

3

2

1

0

1

0 10 20 30 40 50 60

Vertical

stress[M

Pa]

Dam foundationcontact[m]

Verticalstress Spectralanalysis

S

C

S+C

SCSMECME

0.000.200.400.600.801.001.201.401.60

0 10 20 30 40 50 60

Shearst

ress

[MPa]


Shearstress Spectralanalysis

S

C

S+C

SC


The loads taken into account in the elementary method based on strength materials theory (eccentric compression calculus) are presented in Figure 11.

The stress state is in allowable limit, taking into consideration the dynamic character of the seismic stresses.

6. Analysis rtesults by direct time integration

The main hypotheses accepted in the direct time integration method were already presented at point 3.

Some results obtained in this analysis are illustrated in Figures 12 and 13. The component of the baseline corrected accelerogram having a PGA of 0.1g was applied at the finite element mesh bottom on the horizontal upstream - downstream direction.

Fig. 12 - z response time history computed in Direct Time Integration Method due to loads combination of dead

weight+hydrostatic pressure+uplift pressure+accelerogram 0.1g

Fig. 13 - xz response time history computed in the Direct Time Integration Method due to the load combination of

dead weight+ hydrostatic pressure+uplift pressure+accelerogram 0.1g

4500

500

0 1 2 3 4 5 6 7 8 9 101112131415Vertical

stress[K

Pa]

Time[s]

Verticalstresses Directnumericalintegration

Node1Node2

700

300

1300

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15She

arstr

ess[K

Pa]

Time[s]

Shearstresses Directnumericalintegration

Node1

Node2

3.002.502.001.501.000.500.00

0.50

0 10 20 30 40 50 60

Vertical

stress[M

Pa]


Verticalstresses Directnumericalintegration

S+max(C)

S

SMECME


Fig. 14 - z and xz diagrams on dam foundation contact computed in direct time integration at combination of dead weight+hydrostatic pressure+uplift pressure+accelerogram 0.1g compared with z diagram computed by

elementary method

In Figures 12 and 13 are presented the response time history in z (vertical stress) and xz (shear stress) in 5 points located in the profile central section due to dead load + hydrostatic pressure including uplift + accelerogram 0.1g maximum acceleration. The maximum z reaches 378 kPa tension (node 1, upstream toe) and -2766 kPa compression (node 2, downstream toe). The maximum xz directed to upstream reaches 1275 kPa (node 2, downstream toe). The z and xz diagrams on the dam foundation contact computed in direct time integration at the combination of dead weight + hydrostatic pressure + uplift pressure + accelerogram 0.1g compared with the z diagram computed by the elementary method computing stresses + inertia forces computed in compliance with the spectral analysis are given in Figure 14. The corresponding results obtained by both methods are in satisfactory correlations. Some tensile stresses reaching about 400 kPa resulted in the elementary method at upstream dam toe versus 378 kPa in the direct time integration method. Instead at the downstream dam toe the compression computed in the direct time integration method is higher than in the elementary method (-2766 kPa versus -2500 kPa).

7. Comparison of results and concluding remarks

In order to compare the results of the analyses carried out by the response spectrum method and the direct time integration method some representative results are given in Tables 5 and 6, respectively in Figure. 15.

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0 10 20 30 40 50 60

Shearst

ress

[MPa]


ShearstressesXZ Directnumericalintegration

S+max(C)

S

3.503.002.502.001.501.000.500.00

0.50

0 10 20 30 40 50 60

Vertical

stress[M

Pa]


Verticalstresses

DirectnumericalintegrationSpectralanalysis


Fig. 15 - Comparative diagrams of z and xz on dam foundation contact, full reservoir case, computed by spectral analysis, direct time integration and elementary method.

Table 5

Vertical stress - z Node

Method 1 2 3 4 5

Direct numerical integration

min -316 -2766 -414 -1851 0 max 378 -1865 129 -1021 1

Spectral analysis min -741 -3138 -1054 -3239 -10 max 782 -1880 606 -614 15

Table 6

Shear stress - xz Node

Method 1 2 3 4 5

Direct numerical integration min 283 1088 -54 284 0 max 461 1275 245 517 0

Spectral analysis min 90 896 -373 174 -25 max 661 1479 542 912 2

According to Table 5 the maximum z compression from dead weight + full reservoir + 0.1g horizontal earthquake upstream downstream combination is at the downstream dam toe (node 2) reaching -2766 kPa in the direct time integration and -3138 kPa in the spectral analysis (13% difference).

The maximum z tension from the same load combination is at the upstream dam toe (node 1) reaching 378 kPa in the direct time integration and 782 kPa in the spectral analysis (106 % difference). The high difference between the local values may be explained through their small order of magnitude relative to other values that influenced the accuracy of the numerical computation.

According to Table 6 the maximum shear stress xz appears at the downstream dam toe (node 2) reaching 1275 kPa in the direct time integration and 1479 kPa in the spectral analysis (16% difference).

Additionally from Figure 15 one can notice a good correspondence between z and xz stresses computed with the direct time integration and the spectral analysis methods along the dam foundation contact. The values computed by the elementary method are also in satisfactory correlations with their corresponding values obtained in the dynamic analyses.

0.000.200.400.600.801.001.201.401.60

0 10 20 30 40 50 60

Shearst

ress

[MPa]


ShearstressesXZ

Directnumericalintegrat


The comparison between the methods was extended also to the values of the sliding safety coefficient (k) on the dam foundation contact in the full reservoir hypothesis considering the response to the earthquake action by the spectral method and, by the direct time integration respectively.

In comparison the k value resulted by the elementary method of calculus was also considered, in compliance with the relation:

k = f ,, = f (7)

where FA is the dam foundation area;

z - vertical stresses on the dam foundation area calculated with the elementary method xz shear stresses on the dam foundation area calculated with the elementary method V- sum of the vertical loads acting on dam block H - sum of the horizontal loads acting on dam block f = 0.70 friction coefficient to sliding on the dam foundation contact.

The comparison results are summarized in Table 7. Table 7

Comparison between the methods on the values of the sliding safety coefficient (k)

Methods and types of analyses V kN

H kN

k -

Full reservoir foundation on stresses evaluated by FEM ac = 0

567010 307030 1.292

Full reservoir foundation on stresses evaluated by elementary method ac = 0

548490 345110 1.113

Direct time integration ac = 0.1g 500610 353260 0.991 Spectral analysis ac = 0.1g 430400 338900 0.889

Inertia forces evaluated by spectral analysis. Foundation stresses evaluated by elementary

method ac = 0.1g

548490

379930

1.010

According to the results presented in Table 6 the spectral analysis method is conservative compared with the direct time integration methods concerning the dam sliding stability on the foundation.

Based on the data presented above the followings concluding remarks may be formulated:

The aim of the paper was to check the accuracy of the results in the dam seismic response provided by the spectral analysis taking into account some doubt on the performance of this method;

The results in the seismic response of a buttress dam with 73.50 m maximum height computed by the spectral analysis method were compared with those obtained by the direct time integration. In the comparison corresponding results obtained by the elementary method of analysis were also included (linear distribution of normal stresses). The comparison included the stresses in the representative points of the dam body, dam foundation area and the safety coefficient to sliding on the dam foundation contact.

The conclusion is that both the spectral analysis method and the direct time integration method provided close results, acceptable for engineering use. Generally, the differences between the corresponding values are in the range 020%, but locally they can reach even 100% when their order of magnitude is very low relatively to the current values. This may be explained by the accuracy of the numerical computation. As it was expected, the results provided by the spectral analysis method are the conservative ones.


References

[1]. Priscu, R., Popovici, A.( 1970). Optimisation of gravity and buttress dams (in Romanian), Buletinul Stiintific, I.C.B. nr.4.

[2]. Dams in Romania. (2000). Romanian National Committee on Large Dams. Univers Enciclopedic Publishing House, Bucharest.

[3]. ABAQUS 6.11. (2009). Abaqus / CAE Users Manual. United States of America: Abaqus Inc. [4]. Popovici, A. (1978). Calculul structurilor hidrotehnice. Analiza dinamic prin metode numerice (204

pagini) I.C.B. [5]. Popovici, A., Ilinca, C., (Romania), Ayvaz M. T. (Turkey). (2013). The performance of the neural

networks to model some response parameters of a buttress dam to environment actions 9-th ICOLD European Club Symposium, Venice.

[6]. Kashima, T., Kondo, M., Enomura, Y., Sasaki, T. (2014). Effects of reservoir water level and temperature on vibration characteristics of concrete gravity dam. International Symposium on Dams in a Global Environmental Challenges, Bali-Indonesia.

[7]. Meghella, M., Furgani, L. (2014). Nonlinear seismic analysis of dams. International Symposium on Dams in a Global Environmental Challenges, Bali-Indonesia.

[8]. Ito, T., Sasaki, T., Yamaguchi, Y., Annaka, T. (2014).Attenuation relationship of earthquake motion at dam foundation in consideration of the 2011 Tohoku earthquake. International Symposium on Dams in a Global Environmental Challenges, Bali-Indonesia


SIMULATION OF TWO HIGH PRESSURE DISTRIBUTION NETWORK OPERATION IN ONE-NETWORK CONNECTION

SORIN PERJU - Lecturer, PhD, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected] MDLIN MIHAILOVICI - Associated Professor, PhD, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected] IOANA STNESCU - Assistant Professor, PhD, Technical University of Civil Engineering of Bucharest, Faculty of Hydrotechnics, e-mail: [email protected]

Abstract: The programs developed by the water supply system operators in view of metering the branches and reducing the potable water losses from the distribution network pipes lead to the performance reassessment of these networks. As a result the energetic consumption of the pumping stations should meet the accepted limits. An essential role in the evaluation of the operation parameters of the network performance is played by hydraulic modeling, by means of which the network performance simulation can be done in different scenarios. The present article describes the concept of two high-pressure network coupling. These networks are supplied by two re-pumping stations, in which the water flows were drastically reduced due to the present situation.

Keywords: network distribution, pumping stations, numerical model, EPANET

1. Introduction

Designed in different stages and completed on the basis of knowledge, technologies, equipment and materials established for the respective stages, the water distribution network in operation was designed and built in a period of time when the energy cost was a factor of reduced weighting in the exploitation costs. Today, the energy price tends to comply with energy prices of the EU countries and the energy in the potable water price is a considerable weighting factor. Within the distribution networks, the pumping and repumping stations and the house water supply plants are the greatest energy consumers that secure the pressure and the outflow to the user branch, with a major influence in the management of the potable water pumping and distribution process [1]. The progress registered in water pumping engineering, as well as the advanced technology in the control and tracking of the hydraulic system operating parameters, consisting of pumping station distribution network, allowed for exploitation performance, and new operating facilities in the system. The remote control and the automation of the pumping repumping stations are the main elements, that, once implemented, have created a favorable environment for a certain and profound knowledge of the operation parameters, giving the possibility for research studies like: the analysis and balance sheet of energy consumption, the quantities of the pumped and distributed water, the lifetime of the pumps, the system dysfunctions.

In this context, the optimization of the energy costs of the integrated water distribution systems aiming at reducingpower consumption, represents a big issue, that is, a need for energetic and functional optimization of the water distribution system, for both new systems and the operating ones, followed by the elaboration of some hydraulic models and high-performance software for better solving the water distribution systems analysis and design, as well as the measures and solutions for the susbtantial reduction of water loss in the system.

The performance optimization of both the water distribution network and the associated pumping stations, is done by using a modern and adequate mathematical software, thus being possible to establish numerical models on the basis of which the exploitation scenarios in different configurations can be analysed. The results of the analysis and the draw up reports represent the basic elements in the final decision for the rehabilitation and the endowment of the water distribution network and the pumping stations.


2. High pressure water distribution network in Tei-Colentina district

Located in the Northern Eastern side of the town, the district Tei-Colentina utilises a water distribution system developed along time, in stages, according to the area urban extention. Therefore, together with the construction of apartment buildings, it became necessary to work out new water distribution networks, in order to ensure the parameters needed for potable water in the GF+8 GF+10 buildings. Thus, three pumping stations were built, in different stages between the 70s and the 80s : Teiul Doamnei, Petricani and Lacul Tei, located in the areas shown in Fig. 1.

Fig. 1 The location of the pumping stations in Tei-Colentina district

Initially, Teiul Doamnei and Petricani pumping stations were designed and built to operate as classical pumping stations sucking from a series of reservoirs and discharging in a high pressure pipe network, that serves a housing district with a working lift of GF+8 GF+10. Lacul Tei pumping station has an inlet directed in the low pressure network and an outlet directed to the high pressure network associated to this station. The endowment of the stations consists of Romanian pumps models NDS, DN / TN, activated with fixed speed [2]. The control of the working conditions for the 3 pumping stations was done by the sequential start up and shut down of the the pumps, whose control was performed by the operation personnel, that intended to correlate two parameters: the pressure in the outlet and the water table in the inlet reservoirs.

In time, the operating conditions of the pumping stations went through major changes, that led to the modification of the working parameters, in such a way that the parameters of the pumping stations had significant differences compared to the nominal parameters of the pumps, that is, they registered high energetic consumptions.

Thereby, a first step was developed in the 90s, when due to the socio-economic changes, the water consumption was reduced, the pressure in the low pressure network went up, therefore at Petricani and Teiul Doamnei pumping stations, the inlet reservoirs were put out of service, and the pump inlet was directed to the low pressure network. This was the operation pressure that


was used in the re-pumping stations (that varried between 15-25 mCA). Apparently, this way of exploitation seemed to bring an advantage, but in reality, due to the pump parameters, that were Q=360 mc/h and H=60 m, it led to the operation of the partially open outlet valve, so that it should not exceed the maximum accepted pressure in the pipes (6 bar). The pumps worked at low capacity and low efficiency, and registered high energy consumption.

The second major stage in the exploitation condition change was the rehabilitation of the stations based on extensive measurements of the working parameters, by the replacement of the fixed speed pumps with variable speed pumps. Additionally, a study about the high pressure networks served by these repumping stations was performed; the study involved the shut down of Petricani repumping station and the consumers taking over Teiul Doamnei repumping station. Thereby, starting with November 2004 and up to the present, the branches of the high pressure network in the area are being served by the repumping stations Teiul Doamnei and Lacul Tei, stations that are completely rehabilitated and automatic, resulting in substantial energy savings, as it is shown in the diagram of Fig.2, based on the annual registered consumption.

Fig. 2 Energetic consumption evolution in transfer pumping stations

It can be observed that the moment the transfer pump stations were equiped with variable speed pumps (year 2005), the annual energetic consumption was substantial reduced, reaching less than a third from the consumption levels registered when the stations operated with fixed speed pumps.

However, the diagram in Fig. 2, shows that the yearly reduction in energetic consumption follows a continous down trend, for both re-pumping stations. This fact is completely normal as long as:

firstly, immediately after the rehabilitation and automation of the pumps, the water company started a complete program for diminishing water losses in the distribution networks;

secondly, all the branches need to be registered, thus avoiding the lump sum tax consumption.

Compared to the year 2004, the pumped water flows dropped, resulting in only one reduced speed pump operation in the re-pumping station. Moreover, there is a need for new directions for the two neighbouring networks and for the analysis of the possibilities for developing Teiul Doamnei transfer pumping station and the decommission of Lacul Tei pumping station.

3. Hydraulic modeling of the high-pressure water distribution network in Tei-Colentina area (district) For accomplishing the hydraulic model of the high pressure distribution networks and the analysis and assay of the these networks in different operation scenarios, the interface of the


program EPANET 2.0 was used; the program was developed by the Environment Protection Agency, USA, and is designed for hydraulic calculations and for water quality analysis in the distribution networks [3]. EPANET is designed as a research instrument for the correction of the misunderstandings regarding the water flow in the pipe pressure network. It can be used for different applications in analyzing the distribution networks. The phases in developing the network model were:

The precise definition of the network configuration according to the two areas in terms of physical elements (pipeline hubs, pipes, diameters, geodesic head, conjunctions, etc.);

The assignment of the consumption in the conjunction network (the flows were calculated based on the average flows pumped by the two stations in 2012);

The implementation of the distribution network model for the whole study area, by the identification of the conjunctions between the networks (closed valve);

Introducing the characteristic curves of the pumps in Teiul Doamnei station (H=f(Q) and =f(Q)) in the program;

The calibration of the network model, issued on the basis of the energy consumption year 2012;

The performance of the hydraulic calculations in view of optimizing the distribution network on the supposition that Teiul Doamnei station is operating alone.

The final structure of the hydraulic model comprises 236 conjunctions and 283 pipes, an open valve between the two sub-networks and a re-pumping station endowed with variable speed pumps. The simulation of the transfer pumping station that provides the needed flow and pressure in the network junctions was accomplished by using the characteristic curves in the EPANET program (Fig. 3). In order to simulate the hourly consumption variation during a day, the EPANET program allows the user to enter the input of the hourly variation coefficients using the instruction Pattern Editor. The hourly variation flow diagram for each conjunction, related with the average flow, is shown in Fig. 4, according to the variation curve consumption for a large town [4].

Fig. 3- The pump and yield diagrams

Fig. 4 The diagram for the hourly variation consumption


4. Hydraulic calculations and results for optimizing the high-pressure distribution networks in Tei-Colentina district

The first stage in the hydraulic calculation for the high pressure distribution networks for Tei-Colentina area, consisted of the water distribution network simulation considering the decommission of the transfer pumping station Lacul Tei. For that purpose, in the first step of the system operation, we tested the network capacity, by simulating Teiul Doamnei transfer pumping station as a constant level reservoir that continously ensures: a piezometric head of 50 m, to mirror the present operation (5 bars in the pumping station discharge pipe), and the open isolating valve between the two subnetworks. In the exploitation phase where the water flows supplied by Teiul Doamnei pumping station vary between 55 l/s and 151 l/s (Fig. 5), the values recorded for the pressure variation in junction 216 (which belong to the subnetworks served by Lacul Tei repumping station and which is the most hydraulically unfavoured), do not decrease below 47,5 mWc (Fig. 6). This value is high enough to provide the water supply service for an apartment building of GF+10.

Fig. 5 The hourly water flow variation pumped by the Teiul Doamnei repumping station

Fig. 6 The hourly pressure variation in the junctions 2 (Teiul Doamnei RPS) and 216 (apartment building of

GF+10)

The variation of the water flow in the distribution network, supplied at present by Lacul Tei RPS is shown in Fig. 7, where it is determined that these values range between 6,7 l/s and 18 l/s. Considering that the best value for the monthly average flow in Lacul Tei repumping station


was 15,17 l/s in 2012, the tasks of Lacul Tei repumping station can be overtaken by Teiul Doamnei repumping station, without affecting the actual operation service.

Fig. 7 Hourly variation of the flows in the sub-network supplied currently by Lacul Tei RPS

The configuration of the high pressure network and of the distribution in the network conjunctions at 7 AM ( the hour of maximal consumption) is shown in Fig 8, in the hypothesis of Teiul Doamnei repumping station operation so that the piezometric head achieves a pressure of 50 mWc. It can be observed that in no conjunctions the pressure falls below 47 mWc.

Fig. 8 The pressure variation in the Tei-Colentina high pressure network

In the second phase of the hydraulic calculation, in order to find out the energetic consumption in t Teiul Doamnei repumping station and considering the shutdown of Lacul Tei repumping station, we obtained the simulation of the high pressure distribution network for Teiul Doamnei repumping station, endowed with variable speed pumps. This simulates the actual station control, disposed for a 50 mWc pressure daily operating in the outlet pipe. In order to describe this operation step, EPANET program enables the user to enter some basic compilations, so that the variation of the speed pump can be simulated. These basic compilations are meant to maintain 50 mWc in the outlet pipe and to modify the pump speed according to the changes in the hourly consumption, followed by the variation of the characteristic diagram of the pump. The speed was modified by up to 70% from the nominal speed, with a variation step of 2. In this way, 62


courses were written, where the increase and the decrease of the pump speed were described (Fig. 8 represents an example of the first and the last two instructions).

Therefore, having introduced the characteristic diagram H=f(Q) and the yield diagram =f(Q), we equipped the station with 4 pumps. Afterwards, we calibrated the model as to operate, using the Teiul Doamnei repumping station and the high pressure network that it currently supplies, Fig. 9.

Fig. 9 Exploitation scenarios and the configuration of Teiul Doamnei repumping station

Within the exploitation data registered for October, 2012 (the period with the highest energy consumption and the highest pumped flow), the calibration was accomplished and it is shown in Fig.10.

Fig. 10 The input and the pumped flow in Teiul Doamnei RPS and Lacul Tei RPS

Hydraulic calculations were performed within the calibrated model, on the supposition that Teiul Doamnei repumping station supplies the water flow needed for the customers connected to the high pressure distribution network previously serviced by Lacul Tei repumping station. This operation was possible by opening the isolation valve with a diameter Dn of 250 mm located between the two sub-networks.

We obtained results from the simulation of exploiting the network, given that the values of the pumped flow are the same as the ones we had simulated for the station operating like a constant-level reservoir. The results highlighted that the daily hourly variation of the pressure in the most unfavoured conjunction (conjunction 216) ranged between 45.5 mWc and 49.5 mWc, values


sufficient for the water supply of a GF+10 apartment building, as it can be observed in the diagram obtained from simulating the operation of the network.

Fig. 11 - Hourly pressure variation in conjunctions 2 (Teiul Doamnei RPS) and 216 (apartment building GF+10)

The hourly average power demand registered in Teiul Doamnei repumping station in this configuration and with variable speed, is 34,1 kW. It results that the repumping station will have a yearly power demand of 34.1 x 24 x 365 = 298716 kWh, by operating in these conditions. The values for the power demand after the simulation in the given conditions are listed in the diagram below, Fig.12.

Fig. 12 The consumptions for Teiul Doamnei RPS

5 Conclusions

In the study case associated to Tei-Colentina district, the analysis of the optimisation of the distribution network was performed exclusively for the high pressure networks, at present supplied by Teiul Doamnei and Lacul Tei repumping stations.

The concept for elaborating the study regarding the rehabilitation of the high pressure network, started from the need to decrease the power demand, in addition to improving the operating hydraulic parameters of the pumps in Teiul Doamnei repumping station. The study was needed due to the fact that after the station was rehabilitated in 2005 (by replacing the pumps and automating the station), the water company started a project/plan for the water loss decrease. The program was developed along with the one for registering branches resulting in significant decreases of the pumped water flows in the distribution network.


Following this approach in decreasing the water losses and knowing the consumption levels, the pumps in the repumping station work at different parameters compared to the initial ones.

Therefore, a solution implying no suplimentary costs and no changes in operation is the extension of the influence area of Teiul Doamnei repumping station, that could overtake the capacity of Lacul Tei repumping station, thus producing two great advantages:

The pumps in Teiul Doamnei repumping station will work at parameters that are almost equal to the nominal ones, improving the pump capacity;

The shutdown of Lacul Tei repumping station would save 10.376 kWh yearly (if we consider 2012 as a reference year), not taking into account the maintenance costs for the above station.

From the present analysis and based on the simulations accomplished with the EPANET program, it appears that these operating conditions are the most fesable, mainly due to the fact that Teiul Doamnei repumping station has enough capacity, considering that the station is endowed with 4 HS pumps GRUNDFOS (2 fixed speed pumps and 2 variable speed pumps), of which only a single pump is mainly used in the current operation.

The subject of the present article was presented within the Scientific and Technical Conference Water Services and the New Energy Challenges, June, 2013.

References

[1]. Perju S. (2006). The monitoring and hydraulic optimising of the water distribution networks for rehabilitation, PhD Thesis, Bucharest.

[2]. Anton A., Perju S., et.al. (2003). Measurements of the hydroenergetic parameters and network assays for 24 repumping station. Hydraulic Study, Bucharest.

[3]. Rossman, L. EPANET 2 Users Manual. (2000) U.S. Environmental Protection Agency, 600/R-00/057, Cincinnati, OH.

[4]. ***, STAS 1343-66, Hourly variation coefficients for the daily water consumption diagram in the inhabited areas.

[5]. Perju, S., Mihailovici, M. (2013). Optimizarea functionarii retelelor de inalta presiune din zona Tei Colentina, Scientific and Technical Conference Water Services and the new energy challanges.


UPDATE OF THE P100-1 CONCRETE PROVISIONS

VIOREL POPA - Lecturer, PhD, Technical University of Civil Engineering, Faculty Civil Engineering, e-mail: [email protected]

Abstract: In an effort to improve the harmonization of the Romanian design codes with the Eurocodes, the revision of the Seismic Design Code, P100-1, started in April 2010 and ended in September 2013. The main

modelling nr3 2014

Documents

civil engineering bucharest

building services engineering

conical tunnel burner

low emissions burners

conical furnace burner

technical applications

cylindrical tunnel burner

karlowitz effect