ML20151D104

Comments on Westinghouse Arguments Re Plant Steam Generator Tube Rupture.Westinghouse Unconvincingly Explained Incident But Arguments Re Not Expecting Simultaneous Rupture of Many Tubes Seems Reasonable

ML20151D104
Person / Time
Site:	North Anna
Issue date:	03/07/1988
From:	Catton I Advisory Committee on Reactor Safeguards
To:	Ward D Advisory Committee on Reactor Safeguards
References
ACRS-CT-1921, NUDOCS 8804130393
Download: ML20151D104 (6)

Text

/

/e.s'k UNITE D STATES U-/9d/ l

/- . c o NUCLE AR REGUL ATORY COMMISSION

$.' ADVISORY COMMITTEE ON REACTOR SAFEGUARDS

~

g. ' c WASHINGTON, D. C. 20555

..... March 7, 1988 i

gpA 1 0@

MEMORANDUM FOR: David Ward FROM: Ivan Catton SUEJECT: NORTH ANNA SG TUBE VIBRATION ANALYSIS PEFERENCES: 1. H. J. Connors, Jr., "Fluidelastic Vibration of Tube Arrays Excited by Cross Flow," Flow Induced Vibration in Heat Exchangers," ASME, New York, 1970

2. M. J. Pettigrew, J. H. Tromp, and M. Mastorakos, "Vibration of Tube Bundles Subjected to Two-Phase Cross-Flow," Symposium on Flow-Induced Vibrations, Vol. 2, Vibration of Arrays of Cylinders in Cross-Flow, M. P. Paidoussis, ed.,

ASME, New York, pp 251-268 December 1984

3. F. Axisa, B. Villard, R. J. Gibert, and M.A.

Boheas, "Vibration of Tut,e Bundles Subject to Steam-Water Cross Flow: A Comparative Study of Souare and Triangular Pitch Arrays," Pro-ceedings of the Eighth SMIRT Conference, Brussels, August 1985

GENERAL COMMENT

S Westinghouse believes that the North Anna steam generator tube rupture was the result of high cycle fatigue failure. In that the usual thennal hydraulic calculations do not yield cyclic stresses of sufficient amplitude to cause fatigue failure, they argue that fluidelastic in-stabilities occur as well as the usual vortex shedding and turbulent -

pressure oscillations. It is clear from the papers cited and reviewed below that if fluideiastic instabilities occur, the vibration amplitudes -

will be of sufficient amplitude to cause fatigue failure. The question is whether or not their arguments that such an instability occurred are valid. Following the review of the papers cited by Westinghouse, I will address the question.

1 8804130393 880307 19 PDR

C North Anna SG Tube >

Vibration Analysis March 7, 1 5 DOCUMENT REVIEW In response to our request for information, Westinghouse sent the abase 7 three papers to Paul Boehnert. The three papers deal with mechanises i for inducing tube vibration in tubes where cross flow exists. The three l papers deal only with flow nonnal to the tube axis. A summary of what i the three papers contain is given below. In my view, the North Anne :

problem is not one of fluidelastic instability as claimed, because such an instability leads to very large amplitude vibrations straight away and they should have become apparent much earlier in the life of the steam generator. Further, as noted by the authors of the three papers. l our preser. understanding is limited to flows being normal to the tubes. l Little is known about steam generator geometries where the flow may take i on almost any angle relative to the flow from normal to parallel. .

Connors (Ref. 1), studied the tube vibration excitation mechanisms driven by cross flow. He developed a stability criterion for predicting the onset of large amplitude vibrations in single and multiple arrays that takes intn account the interaction between the moving tube and the fluid ficw. His stability criterion was then verified by dynamic stability tests on flexibly mounted tubes. The experiments were carried out in a 30 HP open circuit wind tunnel with an 8x8 inch test section.

The tubes used were designed to have resonant frequencies in the 10-40 Hz or 200 Hz ranges. The tubes were 8 inches long and one inch in diameter. The experiment was conducted with both single and multiple row tube configurations. Connors notes that there are three primary mechanisms causing flow-induced vibration of tube arrays: vortex shedding, turbulence, and fluid elastic excitation. Vortex shedding results in an alternating force as the vortices are shed first from one side of the tube then the other and is only a problem when the departure frequency is approximately ec,ual to the tube natural frequency. For a ,

single tube, the phenomena is well understood. Unfortunately, this is I not the case for multiple tubes where the vortex shedding frequency (sometimes called the Strouhal frequency) is a strong function of the tube spacing and array configuration. Turbulent pressure fluctuations occurring in the wake of a cylinder or carried to a cylinder from an upstream disturbance can be a potent mechanism for tube excitation.

Turbulence will induce vibration in tubes at all flow speeds - although the amplitude will be small at low velocities. Again, the single tube is well understood, with closely packed tube arrays being an area for research.

Fluid elastic excitation occurs as a result of variations in the lift .I and drag coefficients as the angle of attack changes. The angle of attack changes as a direct result of the vibri.tional movement of the tube. As a result, one can see that the vibrations are clearly self excited. In an array of circular tubes, the momentary displacement of one tube effects its neighbors. It is this mechanism that is charac- l terized by a threshold velocity. Connors argues that tubes in single l

! 1

1 a -

l l

2 North Anna SG Tube l Vibration Analysis March 7, 1988 and multi-row arrays experience large amplitude whirling vibrations Whes' the flow velocity exceeds some critical value. These vibrations occur i at the tube's natural frequency causing failure in service. They can only be explained in tems of fluid elastic excitation. The complex-ities described by Connors needs to be kept in mind when one tries to use this work in an assessment of steam generator tube vibration.

Connors' experirent used air as the working fluid which could give results very different from the case of steam and water mixtures. The tube geometries were also different.

Pettigrew, et. al. (Ref 2), presented the results of their experiments on two-phase flow across nomal triangular and normal square tube arrays with a pitch to diareter ratio of 1.47. The two-phase flow working fluid was air-water, based on the argument that steam-water would be .-

"softer" and thus conservative in establishing vibration potential.

This conjecture seens to be contradicted by the French work with steam and water as described below. Their results include measurements of damping, hydrodynamic mass, fluid-elastic instability, and random turbulence excitation.

The experimental apparatus by Pettigrew et al was capable of mass fluxes from 0 to 1000 Kg/s/m**2, steam quality from 0 to 40%, and void frac-tions from 0 to 97'.. The work reported is apparently a small part of their complete program where different tubes, different pitch to dia-meter, and other geometric variations will be investigated. Their objectives are to understand and fomulate flow-induced vibration excitation mechanisms in two-phase flow. As is usually done, they ennsidered three excitation mechanisns: (1) fluidelastic instability, (2) periodic wake or vortex shedding, and (3) vibration response to random turbulent pressure fluctuations. It was found that periodic wake shedding, or, as defined above, vortex shedding, is only significant at very low void fractions.

They found that vibration response is generally greater in the flow direction (drag) then in the normal direction (lift) except near the  ;

fluidelastic instability threshold when the opposite is usually true.

They conjecture that fluid-elastic coupling may be greater in two-phase flow when the tubes vibrate in the cross flow direction (lift direc-tion). Coupling between fluid forces and tube motion is required for fluid elastic instabilities, j The fluidelastic instability was found to be very similar to that found .

in single phase fluids. The critical gap velocity correlated with the expression given by Connors. For random turbulence excitation, the l vibration amplitude was found to be an almost linear function of the  !

mass flux below the fluidelastic instability onset. Large damping i ratios up to 8% were measured at intemediate void fractions between 20%

and 80%. The vibration behavior of the tube bundles was affected by the flow regime in two-phase cross flow.

'.' North Anna SG Tube Vibration Analysis March 7,1988 Axisa,et.al.(Ref.3),studiedtubebundlessubjectedtosteamwater -

cross flows. They found that the domir.6nt mechanism causing tube

• vibration was fluidelastic instability. Vortex shedding was found to be far less efficier.t in two-phase flows than in single phase flows such as were investigated by Connors. Turbulent pressura fluctuations at the tube walls were noted to lead to small amplitude vibration which results in lcrg term fretting wear and fatigue. This work began as a result of the cbservation by C.E.N. Saclay and FRAMATOME that data concerning fluid irduced vibretion in two-phase flows were rather limited.

Axis, et. al., present results on flow induced vibration in square and triangular straight tube bundles with a pitch to diameter ratio of 1.44.

Thcir data shcw that the tube vibration amplitude slowly increases with flow velocity up to some critical ve16 city that is a function of void fraction. The increase in amplitude seen upon reaching the critical .

velocity is precipitous. The critical velocity at which the steep increase in amplitude occurs is easily discerned from the data even though the graphs in the paper are on a very small scale. Above the threshold, the tube response is clearly due to fluidelastic instability.

Mr. tion of tubes in a regior are clearly correlated indicating that the fluid is an active participant in the process. Below the critical velocity, the amplitude is a slowly increasing function of the cross flow. Here the driving force is the turbulent pressure fluctuations.

When one tries to discern the effect of void fraction, the process becomes more complicated. The experiments were conducted with mass qualities ranging from 0.06 up to 0.34. At the higher qualities, tube response increases less regularly with flow velocities. At the highest values Alexis, et. al., found some indication that vortex shedding was contributing to the tube motion. The onset of fluidelastic instabil-ities was, however, still dranatic and overwhelmed all other mechanisms

• that might drive tube vibration. After looking at the data, I find it hard to accept the explanations we were given at the Meeting regarding the North Anna stean generator tube problem. Fluid elastic instabil-itics seem to drive a catastrophic instability, whereas the turbulent pressure fluctuations seem to drive a more ordered process. The data in the paper is admittedly partial and, as a result, my observations are tentative.

Tube damping factors were also measured by Alexis, et. al. The damping measurements produced scattered results. The values were seen to depend on steam-weter quality, flow velocity, direction of tube motion, and on -

the tube under consideration. It was not clear to the authors whether '

or not the scattere.1 results were real, and a function of par 6 meter variation that was not understood, or due to experimental uncertainty.

The measured values ranged from 0.5%, in a nomal triangular array at a quality of 0.34, to 3.4%, in a parallel triangular array at a quality of 0.059. Even so, the measured values were consistently less than those ofPettigrew,et.al.(Ref.2).

C North Anna SG Tube Vibration Analysis M rch 7, lime CONCLUDING REMARKS To detemine whether or not fluidelastic instabilities occur in the upper regions of a U-tube steam generator, one must first establish the magnitude of the threshold velocity for the given geometry and flow characteristi;s. This was not done at the Subcomittee meeting nor is it done in the cited papers. The papers certainly point one in the right direction. The next step should be to do some experiments with prototypical geometries and flow conditions. In the interim, one could make the argument that it is only the cross flow that matters-making the cited papers useful for an analysis. There are reasons, however, to believe that this would be highly approximate.

Given that we are willing to accept that only the flow normal to the tubes is important, one must still calculate the void fraction and liquid and steam velocities through the tubes. This was done by Westinghouse using potential flow theory. Potential flow theory will yield velocities, but their values have no meaning for the problem at hand. More complicated calculations need to be done that treat two-phase flw through rod bundles with phase change. Part of the complex-ity is a result of the water wanting to flow laterally while the steam rises. This is a troublesome calculation for the best of our codes. As a result, it is my opinion that Westinghouse has not demonstrated that they know what caused the steam generator _ tube rupture.

In my view, a more likely cause of the problem is turbulent pressure fluctuations. These lead to fatigue of the tubes in a slow but sure marner. That past therinal hydraulic calculations do not yield forces  ;

that will drive the tubes hard enough, may be a result of not knowing j how to do the calculations in the bend region of the tubes. Given the complex geometry and difficulties in calculating the flow magnitude, one could easily be off by a great deal in estimating the cycle fatigue. As shown by the disagreement in estimates of the damping between the second and third papers, knewing the damping characteristics of a given tube is doubtful. This is compounded by uncertainties in the effectiveness of the antivibration bars.

The Westinghouse cure for the problem is to reduce the recirculation rates in the steam generators. If one knew the threshold value for fluidelastic instabilities, and the cause was such instabilities, then the Westinghouse cure would be effective-providing the velocity through  ;

the tubes is belew the threshold value. On the other hand, if the cause . I is turbulent pressure fluctuations, then the amplitude will, at most, be reduced linearly with recirculation reduction. To be helpful, a signif- i icant change in the recirculation ratio may be required. -

1 l

l

o d North Anna SG Tube Vibration Analysis March 7. 1988 I don't believe Westinghouse has convincingly explained the North Anna incident. On the other hand, their arguments about why one should met expect sinultaneous rupture of many tubes seems reasonable. I do, however, expect that we will see more and more steam generator tube ruptures as more and more tubes reach their fatigue limits. An increas-ing frequency of steam generator tube ruptures is also a concern of the irdustry. The original calculations of potential flow vibrations were based on inadequate knowledge of the thernal hydraulics. It is not' clear that we have the knowledge now. The sensitivity to vibration is a strong function of tube-to-tube and tube-to-shell. clearances., Jn.many.. ,.

cases, internal redifications have been made and their implications are ',. ;

unknown. Our ability.to calculate the flow is very limited, due to the cc plex steam separation process taking place in the upper part of the tube bundle. The tube curvature and placement of antivibration bars .

only complicates the problem. Our knnwledge of the tube damping coeffi-cierts is very poor. It is the view of some that we cannot calculate.it ,

and, as you krow, it has not been reasured. It seems to me that this is ar. area where RES could very effectively contribute to the safety of nuclear power stations by implerenting a program to address some of the areas where we are in ignorance.

s

.