AAR AT MOXOTÓ-TEN YEARS OF MONITORING AND REMEDIAL MEASURES

João Francisco A.Silveira1,Alberto Jorge C.T.Cavalcanti2,Paulo M.D’Amorim2,Evandro G.Wanderley2 ,Luiz Alberto C.Mariz2

Abstract

The concrete structure of the Apolonio Sales Power Plant isundergoing a process of expansive reaction which hasseriously affected the generating units. Between 1988 and1992, three expansion slots were cut in the concrete blockswith the objective of improving the operating conditions.

This paper presents the results of monitoring the turbineand the generator, and analyses the effect of the cutting ofslots on the performance of the concrete structures and thegenerating units.

It is concluded that there was improvement in theperformance of the units to a degree sufficient to maintainthe units in operation, thus postponing the necessity ofcarrying out major repairs.

Introduction

The Apolonio Sales Plant, which is part of the MoxotóHydroelectric scheme, was built between 1972 and 1977 on theleft bank of the São Francisco River, immediately downstreamof the Moxotó tributary. It is equipped with four 110 MWKaplan turbines with 550 m3/s nominal discharge, and beganoperation in 1977.

Since the commissioning phase, the operation of thegenerating units has presented problems which have

1 PROMON Engenharia Ltda. Av. Juscelino Kubistchek 1830, T III, 9°°°°Andar. 04543-900 São Paulo, Brazil. Fax: (011) 827-4924

2 Companhia Hidro Elétrica do São Francisco. Rua Delmiro Gouveia 333.50761-901. Recife, Brazil. Fax: (081) 228-4310

progressively become worse, culminating with the rubbing ofthe turbine blades on the discharge ring.

At the same time, the opening of cracks in the concretestructure, the closing of contraction joints between blocksin the lower part of the structure and the opening of jointsin the upper levels were observed (Cavalcanti,A.J.C.T.,1986).

After an extensive investigation program (Cavalcanti,A.J.C.T., Silveira,J.F.A., Degaspare,J.C.,1991), it wasconcluded that the cause of the irregular behaviour of thestructure and of the generators was the expansion of theconcrete due to alkali-aggregate reaction. This reaction,between the alkalis of the cement and the silica/silicate ofthe aggregates, produces an alkaline gel that expands withthe absorption of water. As a consequence, there is a slowvolumetric expansion of the concrete mass which causesdeformations and eccentricities in the turbine components.

It was possible to keep the turbines operating by carryingout repairs, including complete re-centering by relocatingthe axis about 6mm. To minimize the effects of expansion onthe generating units, the cutting of expansion slots betweenthe concrete blocks was analyzed through a mathematicalmodel. This study showed that cutting slots would cause areduction of the tensile stresses on the stay vanes andwould reduce the ovalling of the circular openings.

Between March 1988 and February 1992 three expansion slotswere cut between the concrete blocks that contain thegenerating units. Each slot had an area of approximately 700m2, and was cut by using an abrasive impregnated steel wiretechnique. To produce a 30 mm cut, six wires with 12 mmspacing were used in each slot (Silveira,J.F.A.,Degaspare,J.C., Cavalcanti,A.J.C.T.,1989).

This was the first time that this method was used with thepurpose of improving the operating conditions in thepowerhouse of a Hydroelectric Power Plant in which concreteexpansion had occurred .

Behaviour of the concrete structures

The displacements in the concrete structures are beingmonitored by multiple rod extensometers installed vertically

and sub-vertically from the internal galleries. Theseinstruments cut across 15 to 20 m of concrete beforepenetrating the foundation rock mass. Inverted pendulums arealso being used as extensometers, having adapted devices tothe vertical wires in order to permit readings of verticaldisplacements, in addition to the horizontal ones.

Before the cutting of the expansion slots between blocks,the strain rate of the concrete blocks observed since 1984had been constant with time, showing no attenuation. Afterthe cutting of the slots a significant reduction wasobserved in the rate of deformation of the concrete, asshown in Table 1.

Table 1 - Expansion strain rate of the concrete

Structure Annual strain rate (microstrain/yr)Before After

Water Intake 110 55Powerhouse 60 46

This attenuation of the concrete strain rate expansion,after the cutting of the expansion slots, is probably due tothe relief in the vertical direction caused by thelongitudinal expansion of the concrete as a consequence ofthe Poisson effect.

In the last years after the cutting of the contractionjoints, the deformations due to expansion have remainedrelatively constant, with annual variations in the order of45 to 50 microstrain, as indicated not only by the multipleextensometers, but also by the inverted pendulums (which arealso used as extensometers).

After a period of about 10 years of observation, themultiple rod extensometers are showing accumulated concreteexpansion values of 480 microstrains for the verticaldirection (EM 3) and between 300 and 400 microstrains forthe sub-vertical direction (EM 2, EM 4, and EM 5), as shownin Figure 1. The longitudinal deformations were determinedthrough horizontal wire extensometers, but the performanceof these instruments was not considered satisfactory.

The jointmeters installed in the contraction joints, inthree different levels of the structure, show the followingdifferential deformations between blocks:differential settlement..........................0,46 mm/yr.joint opening....................................0,19 mm/yr.joint closure....................................0,37 mm/yr.horizontal shearing displacement.................0,28 mm/yr.

The jointmeters installed for monitoring of existing cracksin the concrete wall of the housing of the generator statordo not indicate any significant displacement in the lastyears, which confirms that these cracks are stabilized

Figure 1. Concrete expansion measured by the extensometers

Behaviour of the generating units

Due to the asymmetric distribution of the concrete mass inthe powerhouse, the expansion of the concrete in thevertical direction is greater on its right side, whichcauses tilting of the turbine cover and of the supportbearing, resulting in loss of verticality of the axes.

The lateral constraint to expansion due to the neighboringblocks and to the rock abutments causes the ovalling of thecircular forms contained in the horizontal planes, with anaxis oriented approximately in the upstream-downstreamdirection. This occurs in the discharge ring, the bottomring and in the stay ring, as well as in the turbine pit andin the generator housing. The combined effect results inthe eccentricity of the turbine runner relative to thedischarge ring and of the generator rotor relative to the

stator. Furthermore, both horizontal and vertical relativedisplacements occur between the bottom ring and the outerhead cover.The data discussed below refer to unit 2, with expansionslots on both sides. In the graphs presented in Figures 2through 7 the origin of the abcissas is January of 1977, andrelevant events are indicated, such as re-centering,adjustments of verticality and of the generator air gap andjoint cutting (C1 between Units 2 and 3, C2 between Units 1and 2, and C3 between Units 3 and 4).The Figure 8 presentsthe cross section of the generating unit.

Discharge ring

The eccentricity of the discharge ring is monitored bymeasuring the radial clearance between the discharge ringand the Kaplan blades at the point which coincides with itsaxis. These measurements are made by turning the turbinerunner in 10 degrees increments, and the diametricalclearance is obtained by adding the diametrically opposedradial clearances. Ovalling is defined as the greatestdifference between the diametrical clearances with respectto two orthogonal axes.

Figure 2. Ovalling of the discharge ring

The expansion of the concrete in the powerhouse led to theovalling of the discharge ring with the long axis forming an

angle of 30 degrees with the upstream-downstream direction,measured clockwise. Figure 1 shows the evolution of theovalling process of the discharge ring as well as thediametrical clearance according to the corresponding shorteraxis. A continuous increase of ovalling can be seen fromthe time the first measurement was made in 1981 until thecutting of the first expansion slot in 1988. From then onthere was a reduction in the ovalling process for aboutthree years, after which it again increased.

In the time period prior to the cutting of the slotsovalling increased at an average yearly rate of 0.50mm/year. Between March of 1988 and July of 1990, during theperiod in which the slots were cut, there was a decrease inthe ovalling rate to a - 0.50 mm/year. After the cutting ofthe slots, ovalling increased to a rate of 0.60 mm/year,beginning July of 1990.

Eccentricity of the turbine runner

The eccentricity between the turbine runner and thedischarge ring is calculated as one-half the differencebetween the diametrically opposed radial clearances. Thiseccentricity is greatest in the direction which forms anangle of 45 degrees with the upstream-downstream direction,measured clockwise. Part of the eccentricity results fromthe deformation of the discharge ring, and the other partfrom the relative displacement between the turbine runnerand the discharge ring.

Figure 3. Eccentricity of the turbine runner

Figure 3 shows the eccentricity and the portions relative tothe deformation of the discharge ring and the relativedisplacement in the direction in which eccentricity isgreatest. Between February of 1985 and March of 1987,before the cutting of the slots, eccentricity increased atthe rate of 1.0 mm/year in the downstream-right bankdirection. Between March of 1988 and July of 1990, whichwas the period during which the joints were cut, the averagerate of increase of eccentricity was 0.70 mm/year, in thesame direction. After the cutting of the slots, betweenJuly of 1990 and February of 1992, the average rate was 0.80mm/year. It is interesting to note that the deformation ofthe discharge ring is responsible for an important part ofthe eccentricity (40 to 50%).

Shaft tilting

The generating unit shaft verticality is determined bymeasuring the horizontal displacements with four plumb linesinstalled in two orthogonal vertical planes, with the unitin the stopped reference position.

Figure 4. Shaft tilting

Figure 4 shows the deflection of total verticality in thefourth plane of measurement, located above the turbine guidebearing, with reference to the first plane, located underthe generator rotor. The other measurement planes arelocated above and below the coupling between axis.

Since the beginning of operation in July of 1977 thegenerating unit has already undergone three re-levelingprocesses of the support bearing to correct shaft tilting.The first was carried out in April of 1984 when the turbinewas re-centered, and the others in April of 1987, duringinspection of the unit, and in December of 1989, after thecutting of the joint between units 1 and 2.

Relative vertical displacement of the stay ring

The wicket gates total vertical clearance represents therelative displacement between the bottom ring and the outerhead cover, in the vertical direction. This is obtained bymeasurements made with a thickness measuring gauge, which isintroduced between the vanes and the head cover/bottom ringin the inflow and outflow areas. Total clearance is thencalculated by adding the averages of the readings obtainedin the upper and lower regions of each vane.

Figure 5. Wicket gates total vertical clearances

Figure 5 was constructed by grouping the vanes in fourspecific regions of the turbine where they show similarbehaviour characteristics: upstream/left (vanes 24 to 5),left/downstream (vanes 6 to 11), downstream/right (vanes 12to 17), and right/upstream (vanes 18 to 23).

In this graph the evolution of the clearances in function oftime can be seen. It is observed that in the period betweenthe re-centering of the unit and the beginning of cuttingthe slots in the block between units 2 and 3 (August of 1984to March of 1988) the outer head cover was moving away fromthe bottom ring in the right /upstream/downstream directionat a rate of about 0.06 mm/year. A faster rate is seen onlyin the downstream/right quadrant, at 0.14 mm/year. Thisbehaviour is attributed to a greater volume of concrete inthis region.

Relative horizontal displacement of the stay ring

The difference between the wicket gates upper clearances isused as an indirect measurement of the relative displacementbetween the outer head cover and the bottom ring, in thehorizontal direction. It is obtained through the differencebetween the inflow and outflow upper clearances of each vanein the closed position, multiplied by the ratio between theheight of the distributor and the distance between theclearance measurement points .

Figure 6 was constructed by grouping the averages of themeasurements carried out through time on the vanes containedin the upstream/left, left/downstream, downstream/right andright/upstream quadrants. The graph shows that in the periodbetween installation of the turbine and the last inspectionthe head cover and the bottom ring were displaced in adirection tangent to the shafts of vanes 10 and 21, locatedrespectively in the left/downstream and right/upstreamquadrants, in the upstream-downstream sense, where thegreatest displacements are found.

Figure 6. Wicket gates relative horizontal displacements

Deformation of the generator air gap

The deformation of the air gap is evaluated by measuring thecircularity of the air gap. This is done by manuallyturning the rotor and measuring the distances between theupper and lower parts of a specific pole and the stator atevery 15 degrees.

The maximum deformation index is obtained as follows: themeasurements of the air gap in symmetrical positions areadded for each direction, then all the differences betweenthe sums corresponding to the orthogonal directions arecalculated, and the maximum value of these differences istaken.

The graph in Figure 7 presents the evolution of the maximumdeformation indices during the period between November of1981 and February of 1993. The variation rates changed from2.2 and 2.3 mm/year before the slot cutting, in the upperand lower planes, to 0.3 and 0.5 mm/year after the cutting.The maximum deformation moved from a steady and continuosincreasing state, to a relative stabilization

Figure 7. Generator air gap deformation

Conclusions

The instrumentation (multiple rod extensometers, pendulumsand jointmeters) installed in the concrete structure of thepowerhouse showed the stress relief caused by the cutting ofthe joints. There was an approximate 50 % reduction of theannual expansion rate in the vertical direction at theIntake, and 25 % at the Powerhouse, in the 2.5 years sincethe cutting of the first joints.

The Table 2 summarizes the average yearly rates of themonitored components during the periods before, during andafter the cutting of the slots adjacent to block 2.

The table shows that the rate of the turbine runnereccentricity, of shaft tilting, of the generator air gapdeformation and of the generator rotor eccentricityattenuated with the cutting of the joints. On the otherhand, there was a rate increase in the relative displacementrate between the upper and lower rings of the distributor.

Table 2. Average yearly rates (mm / year)

Type of control Before During AfterOvalling of the discharge ring 0,65 0,30 0,63

Eccentricity of the turbine rotor 1,10 0,77 0,81

Shaft tilting (mm/m/year) 0,05 -0,02 0,03

Relative vertical displacement ofthe stay ring

0,08 -0,06 0,26

Relative horizontal displacement ofthe stay ring (*)

0,40 0,45 0,61

Deformation of the generator airgap

2,20 2,10 0,40

Eccentricity of the generator rotor 0,50 0,20 0,01

(*) Vanes 6 to 11 and 18 to 23

The greater rate of wicket gates tilting is thereforecompensated by the total vertical clearance increasing,which permits the adjustment of the upper clearances of thevanes. This increase in tilting has required adjustments inthe wicket gates links. It is concluded that the cutting ofthe expansion joints mitigated the deformations transmittedto the turbines and were beneficial to the behaviour of thegenerating unit.

All the effects caused by the cutting of the slots areconsistent with the simulations carried out in themathematical model, with the exception of the widening ofthe clearance between the distributor rings in the verticaldirection. This discrepancy can be attributed either tosome deficiency in the mathematical model or to the cuttingof the joints not having produced complete stress relief.

Acknowledgement The authors wish to express their appreciation to CompanhiaHidro Elétrica do São Francisco and PROMON Engenharia forthe permission to publish this paper.

Figure 8. Cross section of the generating unit

A - Kaplan Head L - Combined bearing supportB - Upper Guide Bearing M - Turbine axisC - Upper cross beams N - Outer head coverD - Upper steel bolts O - Intermediate head coverE - Generator Stator P - Stay vanesF - Generator rotor Q - Inner head coverG - Lower steel bolts R - Lower guide bearingH - Generator axis S - Wicket gatesI - Axis coupling T - Bottom ringJ - Combined bearing U - Kaplan bladesK - Combined bearing cross beam V - Discharge ring

References

Cavalcanti,A.J.C.T., "Alkali-aggregate reaction at theMoxotó Dam - Brazil": 7th International Conference onAlkali-Aggregate Reaction, Ottawa, Canada, 1986.

Cavalcanti,A.J.C.T., Silveira,J.F.A., Degaspare,J.C.,"Alkali-Silica Reaction and Remedial Measures at MoxotóPowerplant": XVIII International Congress on Large Dams,Vienna, Austria, 1991.

Cavalcanti,A.J.C.T., Silveira,J.F.A., "AAR at Moxotó GS -Remedial Measures Development and Implementation":International Congress on Concrete Alkali-AggregateReactions in Hydraulic Plants and Dams, Fredericton, Canada,1992.

Silveira,J.F.A., Degaspare,J.C., Cavalcanti,A.J.C.T.," TheOpening of Expansion Joints at the Moxotó Powerhouse toCounteract the Alkali-Silica Reaction": 8th InternationalConference on Alkali-Aggregate Reaction, Kyoto, Japan, 1989.