ML21323A101

Corrosion in Carbon Steel Raw Water Piping TVA 1979

ML21323A101
Person / Time
Site:	Browns Ferry
Issue date:	09/30/1979
From:	Bain W Tennessee Valley Authority
To:	NRC/RES/DRA
Aird, Thomas - 301 415 2442
Shared Package
ML21323A100	List: ML21323A101 ML21323A102
References
Download: ML21323A101 (59)

Text

CORROSION IN CARBON STEEL RAW WATER PIPING

TE NN ESSEE VALLEY AUTHORITY STUDY OF CORROSION

IN CARBON STEEL RAW WATER PIPING SYSTEMS

PART A: Raw Water Pipe Sampling Program

PART B: Pressure Drop in Carbon Steel Raw Water Piping Systems

PART C: Corrosion Inhibitors For Carbon Steel Raw Water Piping Systems

September 1979

Reviewed by

Approved by

SP3O4

TABLE OF CONTENTS

• Pa ge No.

INTRODUCTION AND BACKGROUND. vii

1.0 STUDY PART A - RAW WATER PIPE SAMPLING PROGRAM 1

I.I Approach.... 1 1.2 Anal y sis of Pi p in g Sam p les 1

1.2. I Percent Volume Analysis 1 1.2.2 Change in Pipe Weight 1 1.2.3 Maximum Wall Thinning 1 1.2.4 Deposit Analysis 1

1.3 Data....... 2 1.4 Results and Observations 2

1.4.1 Phase I Sampling Program 2

1.4.1.1 Results 2 1.4.1.2 Recommendations for Future Work Based on Phase I Results 3

I.4.2 Phase II and Phase III Sampling Program 4 1.4.3 Composite Results of the Phase I, Phase II, 4 and Phase III Sampling Programs

1.4.3. 1 Diameter Reduction..... 4 1.4.3.2 Average Pipe Wall Reduction 6 1.4.3.3 Maximum Pipe Wall Thinning. 7

1.5 Conclusions of the Phase I, Phase II 1 and Phase III Sampling Programs 8

1.5.1 Diameter Reduction 8 1.5.2 Average Wall Reduction 8 1.5.3 Maximum Wall Thinning 9

1.6 I mp lementation of Results IO

I.6.1 Average Wall Reduction and Maximum Wall Thinning 10 1.6.2 Average Diameter Reduction 10

1.7 Additional Work Performed as a Result of the Pi p e Sam p lin g Pro g rams

1. 7.1 Pressure Drop Measurements IO I. 7.2 Water Quality in the Holston River IO I. 7.3 Control of Exterior Corrosion 11 I. 7.4 Test of Corrosion Inhibitors... 11

i TABLE OF CONTENTS (Continued)

Pa ge No.

2.0 STUDY PART B - PRESSURE DROP IN CARBON STEEL RAW WATER PIPING SYSTEMS 12

2.1 Introduction and Pu rp ose 12 2.2 Approach....... 12 2.3 Descri p tion of Pressure Dro p Tests 12

2.3.1 Widows Creek Steam Plant - Unit 8 12 2.3.2 Kingston Steam Plant 13 2.3.3 Gallatin Steam Plant 13

2.4 Sa mp le Anal y sis 14

2.4.1 Widows Creek Steam Plant 14 2.4.2 Kingston Steam Plant 14 2.4.3 Gallatin Steam Plant.. 14 2.5 Hazen-Williams Equation 15 2.6 Darc y Equation..... 16 2.7 Comp arison of the Hazen~Williams and Darc y Equations 17 2.8 Extrapolation of Equations to Predict Pressure Drop 2.9 in Pi p in g After 40 Years of Service..... 18 Discussion of Equations and Recommendations 18

3.0 STUDY PART C - CORROSION INHIBITORS FOR CARBON STEEL RAW WATER PIPING SYSTEMS 20

3.1 Approach..... 20 3.2 Cost of Treatment 20 3.3 Test Pro g ram... 20 3.4 Results..... 21 3.5 Conclusions and Recommendations 21

ii LIST OF TABLES AND FIGURES

. Pa ge No.

Table I Phase I Samples 25

Table II Phase II Samples - Widows Creek - Cumberland and John Sevier - Watts Bar 27

Table III Phase III Samples 32

Figure 1 Average Diameter Reduction Versus Service Life for Flowing and Stagnant Samples..... 34

Figure 2 Average Wall Reduction Versus Service Life for Flowing and Stagnant Samples.. 35

Figure 3 Maximum Wall Thinning Versus Service Life for Flowing and Stagnant Samples...... 36

Figure 4 Correlation of Parameters for Use in Hazen-Williams Equation............... 37

Figure 5 Correlation of Parameters for Use in Darcy Equation 38

Figure 6 Comparison of Measured and Predicted Pressure Drop for Widows Creek 3" Line....... 39

Figure 7 Comparison of Measured and Predicted Pressure Drop for Kingston 6" Line...... 40

• Figure 8 Comparison of Measured and Predicted Pressure Drop for Gallatin 8" Line.. 41

Figure 9 Results of Corrosion Inhibitor Test 42

iii

SUMMARY

During preoperational testing of the Emergency Equipment Cooling Water (EECW)

System at Browns Ferry Nuclear Plant during the swnmer of 1976, certain heat exchangers were found to be receiving inadequate cooling water flow due to a buildup of corrosion products on the interior of the carbon steel piping

,- servicing the equipment. This study was undertaken to determine the pervasive ness of this problem in the TVA system and to develop recommended practices to mitigate its effects in the design of future steam plants.

Part A of this report presents the results of a carbon steel raw water piping sampling program to determine the extent of the problem in existing units.

Approximately SO samples were removed from 9 different plants and analyzed by the Central Laboratories of PSC to determine the chemical composition of the buildup, the average pipe inside diameter reduction, the average pipe wall reduction, and the maximum pipe wall thinning. Although large differences were found in some of the above parameters between samples removed from the various plants, the problem was found to be widespread. In fact, difficulty was frequently encountered in locating original pipe samples in many of the plants because much of the carbon steel raw water piping had become unservice able and had been replaced. Although the appearance and consistency of the buildup on the inside wall of the pipe samples varied, iron oxide was virtually always the principle constituent, and the primary mechanism was always corrosion of the steel piping by the aerated raw water with redeposition of the corrosion products onto the inside wall of the pipe in the form of irregular tubercles.

Under each tubercle was found an area of reduced pipe wall thickness. The results of this study indicate that at the end of the 40 year life of a plant, carbon steel raw water piping will experience an average reduction in the inside diameter of 0.40 inch, an average reduction in the pipe wall thickness of 0.065 inch, and a maximum wall thinning of 0.160 inch due to corrosion.

Part B of this report presents the results of pressure drop tests performed in straight sections of carbon steel raw water piping at three TVA steam plants.

These tests were conducted to determine the impact of corrosion products build up on the interior pipe wall.on the flow-passing capability of the pipe. Both the Darcy and Hazen-Williams equations for pressure drop in piping could be made to conform to the measured pressure drop data when the inside diameter was reduced by an appropriate value and when judicious values for surface roughness and "C" were selected for the Darcy and Hazen-Williams formulas, respectively. The results of this study indicate that when calculating the flow-passing capability of carbon steel raw water piping intended for 40-year service, the Darcy equation should be modified such that the diameter used is equal to the nominal pipe I.D. minus 0 : 4 inch, and the relative roughness, e, should be 0.9 inch. The Hazen-Williams equation may also be used if modified such that the diameter used is equal to the nominal pipe I.D. minus 0.8 inch, and a "C" value of 55 is used.

Part C of this report presents the results of an investigation of corrosion inhibitors for carbon steel raw water piping systems. A zinc polyphosphate corrosion inhibitor was selected as the environmentally acceptable corrosion inhibitor most likely to be effective. This inhibitor and sodium bypochlorite

iv were each tested for approximately one year in a small test stand set up at Sequoyah Nuclear Plant. The observed rate of corrosion was initially very high but decreased rapidly. The zinc polyphosphate inhibitor did not signi ficantly reduce corrosion in this test. Sodium hypochlorite appeared to slightly accelerate the corrosion. Cost of the corrosion inhibitor was found to be quite high. Because of high cost and ineffectiveness as demonstrated in the test, corrosion inhibitors are not likely to be a cost-effective means of corrosion control ~xcept possibly in recirculating systems.

The results of this study should have a profound impact on the design of future raw water piping systems, since the pressure drops calculated by methods reco111Dended herein are significantly greater than those which would be calculated by presently recognized standard methods. For example, for a new pipe design velocity of 4 fps, the pressure drop calculated by the methods recommended herein in 100 ft. of straight 3-inch and 8-inch line would be approximately 12 and 4 times, respectively, that calculated by the Hazen-Williams formula with a "C" of 100. Alternately stated, a 3-inch line sized by conventional means to pass 95 gpm will be capable of passing only 25 gpm with the same available pressure drop, or an 8-inch line sized to pa~s 1000 gpm will pass only approxi mately 450 gpm at the end of its 40-year life.

As a result of the Phase I sampling program, revised pressure drop calculations were performed using the Hazen-Williams equation with C=IOO and an assumed diameter reduction of 1 inch. Based on the revised calculations, a number of changes were made in line sizes and pipe materials for plants now under design and construction. However, even those calculations result in predicted pressure drops of only about one-half the values obtained by the methods recommended herein.

Design options to mitigate the effects of corrosion include periodic pipe replacement, increased line sizes, and the selection of corrosion-resistant materials. Economic studies should be conducted on a case-by-case basis to determine which option to select.

The results of this study indicate the need for additional work to facilitate the mitigation of corrosion effects. The following actions are recommended:

1. Carbon steel raw water piping systems for plants now under design and construction should be reanalyzed using the methods recommended herein for calculating pressure drop in order to determine which lines are likely to become inadequate during the plant life.

2. In order to verify presure drop predictions for large diameter pipe, pressure drop testing of large diameter carbon steel pipe should be carried out if a line suitable for testing can be found.

3. Further testing should be conducted at the Sequoyah test stand to determine the detrimental effect of chlorination, the variation with time of the corrosion rate over a longer period, the effectiveness of corrosion resistant materials, and the effectiveness of zinc polyphosphate with a high initial feed rate and reliable injection.

V

4. For the raw cooling water (RCW) and essential raw cooling water (ERCW) systems of future plants, consideration should be given to serving all system components with a closed, corrosion-inhibited water system. The closed system would be cooled by raw water via an intermediate heat exchanger. Such a design would confine the raw water corrosion to a large diameter piping system, where the effects would be less severe.

vi

INTRODUCTION AND BACKGROUND

During preoperational testing of the Emergency Equipment Cooling Water System at Browns Ferry Nuclear Plant during the summer of 1976, it was found that certain heat exchangers were not receiving their design water flow rates.

Sections of the carbon steel piping which supply water to these components were removed by McKelvey (Ref. 1) and were found to have a buildup of material on the interior which impeded the water flow. An analysis of the samples, performed by the Power Service Center Laboratory in Chattanooga, indicated that iron as Fe 2o 3 was the major constituent of the deposit. The buildup of iron was attributed to oxidation of the pipe interior by one or more types of corrosive action; the primary mechanism being the common corrosion of steel piping by aerated river water. In addition to common corrosion, secondary causes of corrosion such as galvanic corrosion, "iron bacteria," and sulfate reducing bacteria were discussed as potential additional contributors to the corrosive attack of the piping interior.

The uncovering of the corrosion product buildup problem at Browns Ferry raised the question of how widespread and severe is the buildup problem. This report

discusses the approach, results, and conclusions of a three-part study of corrosion in carbon steel raw water piping systems.

Part A addresses the results of a sampling program of carbon steel raw water piping from various TVA power plants. This part of the study is an extension of the work performed by McKelvey but addresses both the corrosion product buildup and the effects on pipe wall thickness of the corrosive action. The primary objective of the study in Part A is to explore and better define the corrosion product buildup problem. Data on the average wall reduction and maximum wall thinning will be presented but no recommendations on a design value or on the interpretation of the observed data will be presented. Part B of the study addresses the severity of the corrosion product buildup by discussing the results of pressure drop tests which were performed in various raw water lines. Part C discusses the use of corrosion inhibitors as a means of controlling corrosion in raw water systems and discusses the results of a one-year test at Sequoyah Nuclear Plant of a potential corrosion inhibitor.

vii

1.0 STUDY PART A - RAW WATER PIPE SAMPLING PROGRAM

I.I Approach

Sections of carbon steel raw water piping were removed from existing TVA steam plants in three phases. Phase I, initiated by Ref. 2, covered the removal and analysis of sections of piping from stagnant and flowing lines at the Colbert, Widows Creek, Kingston, John Sevier, and Gallatin Steam Plants. These sites were selected due to their proximity to plants which were currently under design or construction during the sampling period.

Samples from Watts Bar Steam Plant, Browns Ferry Nuclear Plant, and Sequoyah Nuclear Plant were also taken. The results and observations of the Phase I sampling program led to additional Phase II sampling, initiated by Ref. 3, at the Watts Bar, Widows Creek, John Sevier, and Cumberland Steam Plants.

The Phase II sites were selected in an attempt to determine if the amount of buildup is related to length of service. The Phase III samples were removed from the lines at Widows Creek, Kingston, and Gallatin Steam Plants where pressure drop tests were performed as discussed in Part B.

1.2 Anal y sis of Pi p in g Sam p les

The sections of piping removed from the various raw water systems were delivered to the Power Service Center Laboratories in Chattanooga for analysis. The following measurements were used for this study.

1. 2.1 Percent Volume Reduction

A measured length of a piping sample was sealed at one end and filled with water. The volume of water contained in the sample was compared with the original volume as calculated from the nominal dimensions of a new pipe. The percent volume occupied by the deposit represents the average loss in pipe cross-sectional area and can be related to an average decrease in pipe diameter.

1.2.2 Change in Pipe Weight Four-inch lengths of the pipes were split lengthwise, scraped clean, and then acid cleaned in inhibited hydrochloric acid. The sections were weighed and compared with the weight calculated for new pipe of nominal size. The change in pipe weight can be directly related to a change in pipe wall thickness.

1.2.3 Maximum Wall Thinning

A visual inspection of each piping sample was made after cleaning to identify the deepest pit or area of maximum wall thinning. The remaining wall thickness in the deepest pit was measured with a dial micrometer with a 1/32-inch diameter anvil.

1.2.4 Deposit Analysis

The deposit in each sample was scraped out and analyzed for various constituents.

1.3 Data

Observed and calculated data for the piping samples are shown in Tables I, II, and III and for the Phase I, Phase II, and Phase III sampling programs, respectively. The values of average diameter reduction, average wall thickness reduction, and maximum wall thinning were taken from References 4, 5, and 6, respectively. The percent volume analysis values are taken from the Power Service Center Lab Reports which are documented on the tables under the Reference column.

Average diameter reduction, average wall thinning, and maximum wall thinning as a function of years of service are shown on Figures 1, 2, and 3, respectively. Samples from stagnant lines or near stagnant lines are indi cated by an asterisk(*). A least-squares-regression (LSR) of each set of data is shown on each graph.

1.4 Results and Observations

Piping samples were removed from both vertical and horizontal pipelines in water systems at the plants indicated. Piping systems which were.

normally stagnant as well as systems which normally received continuous flow were both sampled.

1.4.1 Phase I Sampling Program

1.4.1.1 Results

The results and conclusions of the Phase I investigation with respect to diameter reduction were documented by Johnson (Ref. 13). The following is a swmnary of the items given by Johnson and additional comments on average pipe wall thinning and maximum pipe wall thinning:

1.4.1.1.1 The primary mechanism by which scale is formed on the interior of pipe walls in raw water systems appears to be corrosion of the steel piping by aerated river water with continuous redeposition of the corrosion products onto the inside pipe wall. The entire process seems to be influenced by sulfate-reducing bacteria.

1.4.1.1.2 Corrosion of raw water piping and the resultant redeposition of corrosion product scale onto the inside of the pipe was found to a significant degree at all plants that were sampled. The degree varied greatly from sample to sample, but generally the scale accumulation appeared to be progressive with age.

1.4.1.1.3 The WC-5 sample illustrates that stainless steel piping components should not be directly connected to mild steel piping components in aerated raw water systems since the mild steel components will then become sacrificial anodes resulting in their severe corrosion.

1.4.1.1.4 In stagnant, continuously pressurized piping systems, the rate of scale accumulation is generally lower than that found in raw water piping*systems where water flows in a continuous or nearly continuous manner, thereby replenishing the oxygen supply which can then induce further corrosion.

1.4.1.1.5 No significant differences were observed in the corroded condition were completely full of raw water. of horizontal versus vertical runs of pipes as long as the pipes

1.4.1.1.6 Sample JS-I had an unusually high value of manganese dioxide (MnO) buildup on the interior. It was speculated that this buildup coutd be due to the presence of manganese reducing bacteria.

1.4.1.1.7 Values of diameter reduction varied from 0.054 inches at a 12.1 year service life to 0.314 inches at a 34.1 year service life (excluding JS-1). Using a linear regression analysis, the forty year design value was found to be 0.34 inches. This value was based on an analysis of only seven samples, however.

1.4.1.1. 8 Values of average wall reduction varied from 0.0102 inches at a 2 year service life to 0.0347 inches at a 34.1 year service life.

Maximum wall thinning ranged from 0.036 inches at a service life of 2 years to 0.122 inches at a service life of 23.8 years.

1.4.1.2 Recommendations for Future Work Based on Phase I Results

Based on the above observations, Johnson made several recommendations for future work and study which led to the initiation of the Phase II and Phase III sampling programs. Following is a summary of some of his recommendations.

A. The least squares regression fit is based on only seven data points.

Additional samples should be taken to increase the data base.

B. The average corrosion product buildup does not represent the full impact on pressure drop in piping. Pressure drop tests should be conducted and samples should be taken from the piping tested to determine the relationship between buildup and pressure drop.

C. The abnormal condition of the JS-I sample should be further investigated. This type of buildup could have serious impact on

the design of Phipps Bend Nuclear Plant.

D. Evaluations of raw water systems at plants under design and con struction should be performed to determine the impact of corroded conditions on systems performance and to develop plans to mitigate the impact.

As a result of recommendation (A), the Phase II sampling program (discussed in the next section) was initiated to increase the amount of available data. Recommendation (B) resulted in the pressure drop Phase III samples (discussed in the next section). A study of the testing described in Part B of this study and the taking of the

water quality of the Holston River was initiated (Ref. 13A) to satisfy recommendation (C). A formal report has not been submitted to EN DES but preliminary results (Ref. 14) indicate that the Holston River water quality has improved and the level of manganese has dropped significantly.

To satisfy recommendation (D), pressure drop calculations for various raw water systems at plants under design and construction were performed assuming a one-inch reduction in the inside diameter due to corrosion product buildup. Although the one-inch diameter reduction was only an estimate of the effects of a 40-year accumulation of corrosion product buildup, it was felt that the values of pressure drop calculated would be more representative than the values calculated using past techniques.

RecoBDDendations on changing out sections of piping to more corrosion,

resistant materials, and increasing the diameter of certain lines were made when deficient flow conditions were identified.

1.4.2 Phase II and Phase III Sampling Programs

The Phase II and Phase III sampling programs will be discussed together since they both were initiated by the Phase I recommendations. Samples from normally stagnant and normally flowing lines were collected at plants of varying ages to establish the effects of age on corrosion product buildup. No requirements on the horizontal or vertical orientation of the samples were established for the Phase II sampling since the Phase I program showed no appreciable differences between samples removed from the two types of lines. Phase III samples were removed from lines where pressure drop tests were performed. The following observations and conclusions were made after combining all the samples from the Phase I, Phase II, and Phase III samples.

1.4.3 Composite Results and Conclusions of the Phase I, Phase II, and Phase III Sampling Programs

1.4.3.1 Diameter Reduction

1.4.3.1.1 From the scatter of data seen on Figure 1, it can be seen that age is not the only parameter which influences corrosion product buildup.

Large variations in buildup can be seen for piping removed from a given site at a given age (WC & WB) and in some cases large varia tions can be seen from samples removed from a single pipeline (WC-21 thru WC-24 and G-2 thru G-5). It was found that the average buildup in 8-inch diameter piping (G-2 thru G-5) and 6-inch diameter piping (K-7 thru K-10) is on the same order of magnitude as the buildup in the 2-inch, 3-inch, and 4-inch lines taken at John Sevier and Widows Creek. Buildup does not appear to be dependent on pipe diameter.

1.4.3.1.2 Large differences in the appearance and consistency of the corrosion product buildup were found. In some cases, more than products of corrosion were found on the pipe interior. At John Sevier, some of the samples were found to have a large amount of manganese deposit.

Sample WC-24 was found to have a higher level of silica than other samples. Most of the samples had a relatively uniform buildup (very rough surface). However, some samples such as K-7 thru K-10 had almost no average buildup but had large, randomly-spaced isolated tubercles. Contrary to the conclusion reached at the end of the Phase I sampling program, data from stagnant (or near stagnant) lines can be seen to fall within the same areas as the data from flowing lines.

1.4.3.1.3 The samples removed from WB appeared to not be any worse than those removed from WC and JS even though the WB piping had been installed approximately 20 years longer. It should be noted that WB was out of service for a significant period of time (10-15 years) and the status of the piping during that period is not known. The effective age of the piping at WB could actually be less than that shown.

1.4.3. 1.4 Samples WC-15, WC-20, WB-8, WB-9, and WB-10 were found to be galvanized lines. These are plotted on Figure 1 but were not used in determining the diameter reduction in galvanized lines is less than that of the the linear regression curve fit. It can be seen, in most cases, that

other samples. Galvanizing appears to reduce the amount of buildup but is not a complete cure. Informal discussions with users of gal vanized piping (Ref. 15) indicate that areas of zinc wear off after a limited period of time leaving the exposed carbon steel piping.

1.4.3.1.5 The purpose of plotting reduction in diameter as a function of age was to extrapolate the data to establish the expected reduction in

pipe inside diameter after 40 years of service. It can be seen from Figure 1 and the above comments that age effect is only one part of the explanation for corrosion product buildup and that the worst data observed actually occurred between 20 and 25 years of service.

Even though age is not the only factor affecting corrosion, the value selected as the 40-year value must be conservative enough to reflect this service life. (TVA has no steam plants which have been operating 40 years.) To establish this 40-year value, the available data was analyzed and extreme cases of diameter reduction discarded where justified.

Rather than selecting a 40-year value which would exceed all measured values of diameter reduction, it was assumed that some piping in any given system will have a diameter reduction sufficiently small to compensate for isolated cases of extremely large diameter reduction. If these isolated areas of large buildup result in deficient flows, the piping in these areas can be replaced. In addition, the selection of a "worst case" is somewhat arbitrary since data from additional sampling could possibly redefine.the "worst case" value.

JS-1 and JS-8 on Figure 1 were both identified as having a high deposit of manganese. Preliminary results of an analysis of the water quality (Ref. 14) indicate the water quality in the Holston River is improving and that the level of manganese in the river had dropped significantly since 1967. Although the memo indicates that results are inconclusive, samples JS-8 and JS-1 should be viewed as not necessarily representing the future corrosive trend of piping systems whose water is supplied by the Holston River.

Widows Creek samples WC-21 thru WC-24 were all taken from the same length of pipe. The average of these four samples, 0.405 inches, is more representative of the overall buildup along the entire length.

When JS-1 and JS-8 are not considered, and when the average.of WC-21 thru WC-24 is used, the maximum decrease in diameter shown on Fig. 1 is 0.405 inch, which corresponds to WC-16 and the average of WC-21 thru WC-24.

The value developed above represents a reasonable upper limit of the measured data and is conservative enough to be used as a 40-year design value. Although the least squares regression indicates increasing buildup with age, it should be noted that the regression coefficient which indicates "goodness of fit" was very low so that the use of the equation for extrapolation purposes is not recommended.

1.4.3.2 Average Pipe Wall Reduction

As stated in the introduction, the primary objective of the study was to identify the extent and severity of the corrosion product buildup in carbon steel raw water piping systems and to determine the impact of this buildup on the flow-passing capability of the system. After the samples were removed and the marked pitting was observed it was decided to perform additional analyses to determine the extent of the average wall reduction caused by the corrosion. It is not the purpose of the study to establish a 40-year design value of this parameter or to discuss the impact of wall reduction on the structural integrity of the system design. Only the data and observations of the data will be given.

1.4.3.2.1 From Figure 2 it can be seen that JS-5, WC-11, JS-7, and JS-6 had the highest values of average wall reduction. Varying degrees of exterior corrosion were noted on all these samples with JS-5 being the worst.

Control of exterior corrosion would allow these values to be discarded.

1.4.3.2.2 Based on the observed data only, average wall reductions up to 0.0625 inches can be expected in piping if exterior corrosion is not controlled.

If exterior corrosion is controlled, average wall reductions up to 0.040 inches can be expected based on the observed data.

1.4.3.2.3 Data from stagnant lines were seen to fall within the same areas as the continuously flowing samples. As with the diameter reduction analysis, large differences in average wall reduction can be seen in samples removed from a given site and even between samples removed from a given pipeline.

1.4.3.2.4 Samples WC-15, WC-20, WB-8, WB-9, and WB-10 were found to have been removed from galvanized lines. The data, shown on Figure 2 but not used in the least squares curve fit, indicate that galvanizing does reduce the wall thinning but does not eliminate the problem. These data must be used with caution since the wall thinning values are calculated from weight loss and the weight of the zinc coating was not included in the calculations.

1.4.3.2.5 The least squares regression curve shown on Figure 2 should not be used for estimating a 40-year design value. It was found that the regression coefficient which indicates the "goodness of fit" of the equation was very low.

1.4.3.3 Maximum Pipe Wall Thinning

As mentioned in the section on average pipe wall reduction, the primary purpose of the study was to address the problem of corrosion product buildup in carbon steel raw water piping systems. Only the data and observations of the data will be presented with regard to the maximum wall thinning.

1. 4.3.3.1 As seen on Figure 3, large variations in maximum wall thinning were found with the variations in data for samples from one site or from

one pipeline even greater than that for the diameter reduction and average wall thinning data. Values of maximum wall thinning up to 0.160 inches were found in the observed data.

1.4.3.3.2 The largest values of maximum wall thinning were seen in some of the 8-inch samples at Gallatin (G-2 and G-3) and the 6-inch samples from Kingston (K-7 and K-10). Tubercles approaching two inches in height were found in these samples and it was found that areas of maximum wall thinning were usually found beneath the large tubercles.

The avera ge diameter reduction of these samples was less than some of the smaller diameter samples due to only isolated cases of the large tubercles.

1.4.3. 3.3 As noted in the comments on the average wall reduction analysis, samples JS-5, JS-6, JS-7, and WC-11 all had varying degrees of

exterior corrosion. This exterior corrosion also affects the maximum wall thinning calculations so that the maximum thinning for these samples would be somewhat reduced if the exterior corrosion had been controlled.

1.4.3.3.4 The data from stagnant lines fell in the same range as the data from the continuously flowing lines.

1.4.3.3.5 Galvanizing appears to be more effective in controlling maximum wall thinning than in controlling average wall reduction or

corrosion product buildup. Much more data from galvanized lines would be necessary before any conclusions could be made however.

The galvanized sample data should be used with caution since the maximum wall thinning values are calculated from the remaining wall thickness measured by a micrometer. The calculated values are underestimates of the actual thinning since the assumed initial wall thickness of the pipe does not account for the additional wall thickness due to the zinc coating. This results in negative values of maximum wall thinning for some samples.

WB-8 was found to be a unique galvanized sample. Removing the insulation from the piping exterior to obtain the sample revealed a 0.4-inch diameter hole rusted through the pipe wall. It is speculated that the zinc coating was defective at this point exposing the carbon steel to the stagnant water in the line. The corrosive attack was accelerated due to the small area of carbon steel pipe exposed to the full volume of water in the galvanized line.

1.4.3.3.6 The least squares regression curve shown on Figure 3 should not be used for estimating a 40-year design value. It was found that the regression coefficient which indicat~s the goodness of fit" of the equation was very low for this case.

1.5 Conclusions of the Phase I, Phase II, and Phase III Samp lin g Pro g rams

1.5.1 Diameter Reduction

It was found that many more factors other than age affect the corrosion product buildup on the interior of carbon steel raw water piping. Large differences in buildup can be found between samples of piping of the same age removed from a particular site. Substantial differences can even be seen between samples of piping removed from a single pipeline.

The appearance and consistency of buildup can vary substantially also.

Some samples had a relatively uniform buildup over the entire surface whereas other samples had only a slight buildup with large, randomly spaced tubercles.

Some of the samples removed from John Sevier were found to have high levels of manganese in the deposit. At Widows Creek, high levels of silicon were found in some of the samples. Samples of piping were removed from both vertical and horizontal sections of piping from systems which were normally stagnant and from systems which normally had a continuous flow rate. Different diameters were also sampled. The buildup appeared to not be dependent on orientation in the pipeline, flow condition, or diameter. Galvanized samples appeared to have a slight reduction in buildup but the data base was not extensive enough to draw any conclusions.

Since TVA does not have any steam plants which have been in service for 40 years, and due to the large scatter in the data which was observed, an approach was developed to estimate a 40-year design value from the data which was available. Samples which were found to have high levels of buildup were discarded where justified. The largest value of diameter reduction remaining after discarding all possible data, 0.40 inches, was selected as the 40-year design value. Although the approach is somewhat speculative, it is felt that the result is conservative enough to be a design value. There may be some isolated instances where buildup will be so excessive that pipe replacement may be required. The impact of this potential pipe replacement in nuclear safety-related systems would require evaluation.

1.5.2 Average Wall Reduction

The scope of the study did not include the prediction of a 40-year design value of average wall reduction or the impact on system design of the reduced wall thickness. The purpose is only to report the observed data and to make any comments regarding trends or peculiarities.

It was found that the samples which had large values of average wall are considered, average wall reductions reaching 0.0625 inches were seen. reduction also had varying degrees of exterior corrosion. If all samples

Discarding samples with obvious extreme exterior corrosion drops the maximum value of the average wall reduction to 0.040 inches.

Large differences in average wall thickness were found between samples removed from a particular site and even between samples removed from a particular pipeline. The average wall reduction did not appear to be a function of pipe diameter or the flow condition of the piping from which the sample was removed. From the limited data available, galvanizing was found to reduce the wall thinning problem. It was speculated that the zinc coating initially protects the piping against corrosion but as the piping is used, the coating is gradually removed which exposes the carbon steel to the corrosive attack of the water.

1.5.3 Maximum Wall Thinning

The scope of the study was limited to only reporting the observed data on maximum wall thinning and making comments on trends and peculiarities.

Prediction of a design value or the implications of the maximum wall thinning with regard to system design were not addressed.

Values of maximum wall thinning up to 0.160 inches were found in the observed data. No trends were identified with regard to the flowing condition of the piping system, i.e., flowing or stagnant.

Areas of maximum wall thinning were usually found beneath large tubercles.

Tubercles approaching 2 inches in height were found in some 8-inch diameter piping samples from Gallatin and some 6-inch diameter piping from Kingston.

The largest values of maximum wall thinning were found in these samples indicating that maximum wall thinning is a function of tubercle size.

The maximum tubercle height possible in piping is physically dependent on diameter so that the maximum wall thinning in larger diameter piping has a probability of being larger than in smaller diameter piping. It was noted that the average diameter reduction in the larger diameter piping was less than that found for smaller diameter piping due to only isolated cases of the large tubercles.

Some of the samples found to have large values of maximum wall thinning were also found to have varying degrees of exterior corrosion. The calculation of maximum wall thinning is affected by the exterior corrosion so that these data could be discarded if the exterior corrosion were controlled.

Based on the small number of samples available, galvanizing was found to have a beneficial effect in controlling the maximum wall thinning in most cases.

1.6 Implementation of Results

1.6.1 Average Wall Reduction and Maximum Wall Thinning

The results _regarding the average wall reduction and maximum wall thinning have been given to CEB, MDB, and each of the nuclear design projects within EN DES for their use in performing seismic and pressure calcula tions. The data were incorporated into design criteria documents as a characterization of corrosion to aid the design engineer in establishing an appropriate corrosion allowance for a particular calculation. (At the time the data was distributed, the largest value of maximum wall thinning found was only 0.130 inches.)

1.6.2 Average Diameter Reduction

The findings of the Phase I investigation initiated the reworking of pressure drop calculations for various raw water systems in plants under design and construction. A 1-inch diameter reduction was assumed in all carbon steel piping as an estimate of the overall effect of the corrosion product buildup on pressure drop. Recommendations on changing out piping to more corrosion-resistant materials and increasing the diameter of certain lines were made when deficient flow rates were identified.

The 40-year value of diameter reduction found as a result of the Phases I, II, and III investigations was used in developing equations to predict pressure drop in carbon steel raw water piping. This is discussed in more detail in Section B of this report.

1.7 Additional Work Performed as a Result of the Pi p e Sam p lin g Pro g ram

1.7.1 Pressure Drop Measurements

In order to better predict the effects of corrosion product buildup on pressure drop in carbon steel raw water piping, pressure drop tests were performed at the Widows Creek, Kingston, and Gallatin Steam Plants.

The data was correlated with both the Darcy equation and the Hazen-Williams equation for pressure drop. Both equations were found to favorably agree with the data when appropriate values of input parameters were assumed.

For the Darcy equation, it was recommended that an absolute roughness equal to 0.9-inch in combination with a diameter reduction equal to 0.4-inch be used to predict pressure drop in piping which has been in service for 40 years. For the Hazen-Williams equation, it was recommended that the 40-year value of pressure drop be obtained by using a diameter reduction of 0.8-inch in combination with a C-factor equal to 55. A more complete description of the test results and predictive equations which were developed from the test results are described in Section B of this report.

1.7.2 Water Quality in the Holston River

Due to the large amounts of manganese found in the deposits in some of the samples removed from John Sevier Steam Plant, a study was initiated (Ref. 13A) to determine the source of the manganese and to determine if

the same type of phenomena will be expected to occur at the Phipps Bend Nuclear Plant which also is supplied water from the Holston River.

Although a final report is not yet available, preliminary findings (Ref. 14) indicate that the water quality in the Holston River is improving and that the manganese level has dropped significantly since 1967. Results are inconclusive but the samples removed from John Sevier which had high manganese levels should be viewed as not necessarily representing the future corrosive trend of the Holston River.

1.7.3 Control of Exterior Corrosion

Due to the large amounts of exterior corrosion seen on some of the piping samples, an investigation was ma<le into the types of exterior insulation normally used on raw water lines (Ref. 16). It was found that piping is insulated to prevent sweating whenever dripping condensation would cause damage or would result in a hazard to plant personnel. Eight-pound mineral wool with an aluminum jacket has been specified in the past for areas where the insulation has a chance of being destroyed. Three-pound fiberglas or foamed plastic is used on ceiling drains, pipe chases, and other locations which are normally out of high traffic areas.

The samples removed from John Sevier were found to have been removed from a section of piping insulated with mineral wool. However, the piping was in a high moisture area and was periodically exposed to a

~ water spray used for washdown purposes. It was speculated that the spray of water caused the exterior corrosion problem rather than the normally occurring condensation.

TVA has changed the antisweat insulation from mineral wool to a foamed plastic insulation developed specifically to prevent sweating. The foamed plastic has a closed cell structure which stops the passage of air and water vapor, thereby not requiring a separate vapor barrier.

This type of insulation was first specified for piping located in the Sequoyah pumping station where excessive sweating was a problem, and will be specified for use in a forthcoming design guide on insulation.

It appears that exterior corrosion will be controlled at all plants after Sequoyah. The impact of the occurrence of exterior corrosion at plants prior to Sequoyah.has not been evaluated.

1.7.4 Test of Corrosion Inhibitors

One possible means of controlling corrosion in raw water piping is through the use of a corrosion inhibitor. Zinc polyphosphate, a corrosion inhibitor distributed by Calgon under the trade name C-39, was used in a I-year test at the Sequoyah Nuclear Plant. The inhibitor was injected into a I-inch diameter carbon steel line carrying raw water at approxi mately 6 feet per second. Sodium hypochlorite, the chemical to be used in controlling Asiatic clams, was injected into a line identical to the line being injected with the inhibitor. Raw water was allowed to flow in a third carbon steel raw water line as a control.

After one year of testing, there was no significant reduction in the corrosion product buildup due to the corrosion inhibitor and there was an increase due to chlorination. The Sequoyah test stand will be discussed in more detail in Section C of this report.

2.0 STUDY PART B - PRESSURE DROP IN CARBON STEEL RAW WATER PIPING SAMPLES

2.1 Introduction and Pu rp ose

Johnson (Ref. 13) recoomended at the completion of the Phase I sampling program (discussed in Part A) that pressure drop testing be performed in carbon steel raw water lines to evaluate the full impact of the corrosion product buildup. The pressure drop will be increased as a result of the decrease in the average diameter of the piping due to the buildup and as a result of the increase in roughness due to the irregular nature of the buildup. The purpose of Part B of the study is to document and present the results of pressure drop tests which were performed to evaluate the full effect of the corrosion product buildup and to present predictive methods of estimating pressure drop in carbon steel raw water piping after 40 years of service. The complete study of the pressure drop testing is contained in References 17 and 17A.

2.2 Approach

In order to evaluate the effects of corrosion product buildup on pressure rmed at the Widows Creek, Kingston, and Gallatin drop, tests were perfo.

Steam Plants. The sites were selected to cover a range of ages as well as a variety of water sources. All tests were made on straight lengths of pipe to avoid consideration of bends. Tees were included in some of the piping systems tested but the pressure drop across the tee was neglected since the run of the tee was always in line with the test flow and the lateral branch was always closed. Efforts were made to test lines which had typical conditions of service such as velocity. However, this was often impossible because the more typical lines had already been replaced. Samples removed from each test line were analyzed by the PSC Lab to determine the percent volume reduction of the pipe interior due to the corrosion product buildup. The corresponding diameter reduction for each test line was then used with the pressure drop test data to develop appropriate equations for predicting pressure drop. The Hazen-Williams and Darcy equations for pressure drop were considered with each set of data being treated separately and then analyzed to establish a correlation to the other sets of data. Predictive equations were then formulated to predict pressure drop in carbon steel raw water piping after 40 years of service.

2.3 Descri ption of Pressure Drop Tests

Following are short descriptions of the lines tested at each plant:

2.3.1 Widows Creek Steam Plant - Unit 8

Tests were performed on a 3-inch diameter cooling water line to the pulverizers. The pipe was a parallel header to the pulverizers which was not used continuously and had been in service for approximately 13.75 years. The line was divided into 4 sections as shown below.

5

--0

~~:=r 6--+-l-20 -, -9. -, -4+---2-8'----;ii--2-8'----it-,-~-+

Pressure drop tests were performed on sections 1-2, 2-3, 3-4, 4-5, and 1-5. Samples of pipe approximately 2-1/2 feet in length (samples WC-21 thru WC-24) were removed (from sections 1-2, 2-3, 3-4, and 4-5, respectively) for analysis.

2.3.2 Kingston Steam Plant

Tests were performed on a 2-inch line and a 4-inch line which were used to supply water to clean the condensers. The 2-inch line was divided into two sections with the first section being 68'3" in length (Unit 8)'

and the second section being 68'4" in length (Unit 9). The 4-inch diameter line was 78'1" in length (Units 7-8). Both the 2-inch and 4-inch lines had been installed for 12 years. Tests were performed on each of the 2-inch line sections and on the single 4-inch section. One sample approximately 3 feet in length was removed from each section of the 2-inch line.* Three samples approximately 2-1/2 feet in length were removed from the 4-inch line.

After performing the tests on the two line sizes and removing the samples, it was found from conversations with plant personnel that the velocity in the 2-inch line approached 22 feet per second when used and the velocity in the 4-inch line approached 10 feet per second when used. There was no buildup seen in the 2-inch line and only a slight buildup in the 4-inch line. Due to the peculiarities of these systems, the 2-inch line and 4-inch line pressure drop data were not used in the data evaluation.

One of the 2-inch sections was saved however (K-3), and the three sections from the 4-inch line (K-4, K-5, and K-6) were sent to PSC Lab for analysis.

Pressure drop tests were also made in a Kingston Unit 7, 6-inch diameter fire protection line. This line was the only vertical line tested and was 65 feet in length. The pipe had been in service approximately 23 years. Four samples (K-7 thru K-10) ranging from 2 to 2-1/2 feet in length were removed for analysis.

2.3.3 Gallatin Steam Plant - Units 3-4

Pressure drop tests were performed on two sections of an 8-inch diameter fire protection line at Gallatin Steam Plant. The first section of piping was 89 feet in length and the second section of piping was 93 feet in length. The pipe had been in service 19-1/2 years. Two, 2-feet-long samples (G-2 thru G-5) were removed from each section of piping for analysis.

An orifice inserted in a length of new piping was installed in each of the piping systems to measure flow rate. The orifice was installed adjacent to the sections of piping where pressure drop measurements were taken.

Taps were installed in the lines to allow pressure drop measurements to be made and mercury manometers were used to measure the pressure drops across the orifice and each section of piping.

2.4 Sa mp le Anal y sis

2.4.1 Widows Creek Steam Plant

It was found that the samples removed from the 3-inch test line at Widows Creek had a substantial amount of buildup on the interior which was comprised of iron oxide and silicon oxide. The diameter reduction of each sample (see Part A for additional data) removed from the test line was found to be the following:

Pi p in g Section Sam p le ID Diameter Reduction, In

1-2 WC-21 0.259 2-3 WC-22 0.440 3-4 WC-23 0.326 4-5 WC-24 0.594 Average of Sections 1-5 0.405 Average of Sections 2-5 0.453

2.4.2 Kingston Steam Plant

The buildup in the samples removed from the 6-inch line at Kingston was found to have only a small amount of uniform buildup but had very large tubercles (some approaching 2 inches in height) randomly spaced. The diameter reduction of each sample was found to be the following:

Sa mp le ID Diameter Reduction, In

K-7 o. 169 K-8 0.098 K-9 0.101 K-10 0.163 Average 0.133

2.4.3 Gallatin Steam Plant

The 8-inch diameter samples from Gallatin were found to have a more uniform buildup than the Kingston 6-inch line but also had large, randomly-spaced tubercles. The diameter reduction of each sample was found to be the following:

Pip ing Section Sample ID Diameter Reduction, In

A G-2 0.227 A G-3 0.313 B G-4 0.280 B G-5 0.359 Average Section A 0.270 Average Section B 0.320 Average Sections A and B 0.295

2.5 Hazen-Williams Equation

The Hazen-Williams equation (Ref. 18) can be written in the form (1)

where h,. = head loss in feet per 100 feet of pipe

~=pipe inside diameter in inches C = roughness factor Q = flow rate in gpm

A least squared curve fit of the form

where a 1 is a constant, was obtained for each set of data. Equations (1) and (2) were set equal to solve ford, (3)

For each set of data (a unique value of a 1 ), a table of (C,d) values was generated which will satisfy Equation (3).

Several figures were generated in an attempt to find a correlation between diameter reduction and C. The development of the best correlation is discussed here. Values of C were assumed and corresponding values of dCALC were calculated for a given test using Equation (3). A dimensionless parameter, d*, was defined for use in correlating the above calculated value of d with the measured value of diameter reduction.

d* - Calculated Diameter Reduction Measured Diameter Reduction... (4)

where ~OM= nominal inside diameter of new pipe, in

~dc~c = d = calculated inside diameter of pipe using Equation (3)

~AS diameter reduction corresponding to the percent volume reduction measured by the PSC Lab.

d* as a function of C for all the pressure drop tests (including the 4-inch line at Kingston) is shown on Figure 4. Excluding the 4-inch line at Kingston as discussed in Section 2.3.2, all data is enclosed in the area indicated by the dashed lines. The smallest variation of diameter reduction/

measured diameter reduction occurs at a value of C of 55 (or 56) at a value of d* equal to 2.

Using a value of C equal to 55 and a diameter reduction equal to twice that measured, the Hazen-Williams equation becomes

= (d_ _ 2 ~d- )4.8655 0.63 Ql. 85 (5)

~OM ' -MEAS

2.6 Darc y Equation

The Darcy equation (Ref. 19) can be written (for a pipe length of 100 ft) in the form

(6) where h1, Q, and dare the same as defined previously and f is the friction factor. A least squares curve fit of the form

(7) where a 2 is a constant, was obtained for each set of data.

Setting Equations (6) and (7) equal and solving for the friction factor, f, results in f = a2d 5 (8) 3.11 Moody in Ref. 20 presented an expression for the friction factor in fully rough flow as

(9) where e is the absolute roughness of the pipe interior expressed in inches.

(Full rough flow is almost certain to exist at design flow in old, corroded piping.) This equation can be rearranged to

e/d = 3.7/(10 1/2-ffi... (10)

For each pressure drop test (a unique value of a 2 ) values off were calculated for different assumed inside diameters using Equation (8).

Equation (10).was used to calculate t/d. Using the assumed values of d, the corresponding values of & were calculated.

In the same manner as was used for the Hazen-Williams equation, various curves were generated in an attempt to establish a correlation based on the pressure drop tests. The best correlation was found from a plot of

& vs. d*, shown in Figure 5. Excluding the 4-inch line at Kingston, as discussed in Section 2.3.3, all the data is enclosed by the dashed lines.

The smallest variation in e appears to occur at a value of d* equal to 1.0 where & = 0.9 inches. Note that if the data from Section 3-4 of the Widows Creek 3-inch line could be discarded, the agreement would be even better.

Using a value of & = 0.9 inches and a calculated diameter reduction equal to the masured value of diameter reduction (d* = 1), the Darcy equation becomes

(11)

(12)

2.7 Co mp arison of Results Obtained from the Hazen-Williams and Darc y Equations

Figures 6, 7, and 8 show the raw data taken for various pressure drop tests along with curves representing different methods of calculating pressure drop. The curve labelled "Hazen-Williams, C=lOO" represents the prediction method which has been used by EN DES in the past and currently is recommended for use in a Design Guide for pressure drop calculations (Ref. 21). The.

curve labelled "Hazen-Williams, C=lOO, l" Reduction in Dia." represents the pressure drop predictions based on interim recommendations of the Phase I Pipe Sampling Program (see Part A). The predictions of pressure drop using the equations derived in Sections 2.5 and 2.6 are shown as the "Proposed Hazen-Williams" and "Proposed Darcy" curves, respectively.

Note that on all three figures, the predictions of pressure drop using the Hazen-Williams equation with C=lOO and a 1-inch reduction in diameter yields values substantially less than those measured. The Hazen-Williams equation with C=lOO and no diameter reduction grossly underestimates the measured values.

Two approaches were taken in an attempt to identify whether the Hazen Williams or Darcy equation is the more approiriate equation to use. A least squares curve fit of the form~= a Q was performed for each set of data to determine whether "x" was close~ to 1.85 (Hazen-Williams) or 2 (Darcy). The calculated exponents of Q ranged from 1. 707 to 2.12 with an average value of 1.95 and a standard deviation of 0.13.

The second approach used *in evaluating the two equations was to actually compute percent difference between the predicted values and the measured values. In comparing the data from the pressure drop tests performed on Section 1-5 of the Widows Creek 3-inch line, the Kingston 6-inch line, and the Gallatin 8-inch line, it was found that the average percent difference between the pressure drop values predicted by the proposed Hazen-Williams equation and the measured values was 8.0 percent with a standard deviation of 11. 7 percent. Using the same measured data, the average percent difference between the pressure drop values predicted by the proposed Darcy equation and the measured values was 1.6 percent with a standard deviation of 10.0 percent.

On the surface, both approaches appear to indicate slightly better agree ment with the Darcy equation. Although the average exponent calculated by curve fits appears to imply the data is better suited to the Darcy equation, the variation in the individual values of the calculated exponents (1.707 to 2.12) is large enough to make the results inconclusive.

In addition, the slightly smaller percent difference values for the Darcy equation are also inconclusive. The selection of the parameter values for the two equations (e=0.9 for Darcy and C=55 for Hazen-Williams) was based on a visual inspection of the curves which were generated. The percent differences calculated above would change substantially if only a small change in e or C were made. For example, if a C of 56 had been used rather than a C of 55, the values predicted by the Hazen-Williams equation would drop by 3.3 percent. The values selected fore and C do not necessarily represent the absolute optimum values which would give the best possible agreement with the test data, but are reasonable values based on observation. Both the Hazen-Williams and Darcy equations

agree favorably with the data and predict values of pressure drop much greater than that calculated by past practice. The use of either euqation would yield results satisfactory for most engineering calculations.

2.8 Extra polation of Equations to Predict Pressure Dro p in Carbon Steel Raw Water-Pi p in g After 40 Years of Service

For the Hazen-Williams equation, it was found that the pressure drop in piping is best predicted by using a C of 55 and using a pipe inside diameter equal to the nominal inside diameter reduced by twice the measured diameter reduction caused by interior buildup. The Darcy equation for predicting pressure drop was based on an e of 0.9 inches and using a pipe inside diameter equal to the nominal inside diameter reduced by the measured diameter reduction. Thus the values of C and£ were found to be independent of both age and pipe diameter. Extrapolating the equations to predict pressure drop in piping after 40 years of service reduces to establishing a value of diameter reduction expected after 40 years of service.

In Part A of this report, a 40-year design value of diameter reduction equal to 0.40 inches was presented. It was mentioned that this value is a reasonable estimate of what will be expected in piping after 40 years of service but does not represent a maximum buildup. Some isolated cases of buildup more severe than this may occur resulting in flow defi ciencies to some equipment. Pipe replacement may be required in these cases and the impact on the design of nuclear safety-related systems of this potential replacement must be considered.

The resultant Hazen-Williams equation to calculate pressure drop (head loss per 100 feet of pipe) in piping after 40 years of service is

~ = 0.63 Q1.s 5 (~OM_ 0 _ (13) 8 )4.8655

The resultant Darcy equation to calculate pressure drop (head loss per 100 feet of pipe) in piping after 40 years of service is h 1 = 3.11 f Q /(~OM - 0.4). 2

• 5 (14)

where f = [2 log 10 [(4.l)(~OM - 0.4))) -2 (15)

2.9 Discussion of Eq uations and Recommendations

At reasonable velocities, the Hazen-Williams equation consistently predicts values of pressure drop greater than the Darcy equation when Equations (13) and (14) are used. This is apparently because the large diameter reduction projected after 40 years is doubled in Equation (13) but not doubled in Equation (14). The equations reverse in severity at higher flow rates but as pipe diameter increases this transition velocity also increases so that in large diameter piping with practical velocities, the Hazen-Williams equation always predicts higher values of pressure drop than the Darcy equation.

Note that as diameter increases, the values of e/d and f [from Equation (15)] decrease. As a result, the Darcy head loss prediction decreases more rapidly than would be expected on the basis. of diameter alone. This is in contrast to the constant C factor incorporated in the Hazen-Williams equation. No pressure drop data or sample analysis has been performed on large diameter piping (greater than 8 inches). Thus, proving which method is more appropriate for large diameter piping is not possible at this time.

It is reco111Dended that testing of large diameter lines be performed to determine which equation is more applicable, provided that lines suitable for testing can be found.

Therefore, for more conservative calculations, the Hazen-Williams equation should be used. For less conservative calculations (but still much more severe than past practice) the Darcy equation should be used.

Using Equations (13) and (14) for piping with a nominal inside diameter of 2.5 inches and below results in extremely high predictions of pressure drop at flow rates normally used in pipelines of these sizes. Although the equations can be used for these pipe diameters, it is recommended that either corrosion resistant material or oversized lines be used.

3.0 STUDY PART C - CORROSION INHIBITORS FOR CARBON STEEL RAW WATER PIPING SYSTEMS

3.1 Approach

Use of a corrosion inhibitor is a possible alternative to pipe replacement, use of oversized pipe, and use of corrosion resistant material as a way of handling corrosion problems. In order to evaluate the use of corrosion inhibitors, it was necessary to determine the cost and the effectiveness of potential corrosion inhibitors.

Potential corrosion inhibitors were discussed with two suppliers (Calgon and Nalco), a water treatment consultant (Sheppard T. Powell Associates),

and the Plant Engineering Branch of Power Production. Zinc polyphosphate was identified as the type of corrosion inhibitor most likely to be both environmentally acceptable and cost effective in our applications. The supplier's recommended feed rate was 1.5 to 2.0 ppm, with a possibility that 4 ppm might be required for effective corrosion control in once through systems.

Certain types of bacteria can cause accelerated corrosion. Evidence of these bacteria has been found in several samples taken in Part A. Chlori nation could eliminate the harmful bacteria and thereby reduce the extent of the corrosion problem. However, under certain conditions the presence of chlorine can itself lead to increased corrosion. Several raw water systems are to be chlorinated for reasons other than corrosion, Because of these factors, chlorination was tested as a potential corrosion inhibitor along with the zinc polyphosphate.

3.2 Cost of Treatment

The present worth cost of the zinc polyphosphate corrosion inhibitor alone (excluding the cost of equipment, maintenance, and other operating costs) was estimated on the basis of a 40-year plant life assuming 5 percent inflation and 8 percent interest. For a 2 ppm feed rate, the present worth cost is approximately $150 per gpm treated. The present worth cost of chemicals to treat a 30,000 gpm flow, for example, would be on the order of $4.5 million. Because of this high cost and lack of confidence in the effectiveness of corrosion inhibitors, no attempt was made to determine the cost of equipment, maintenance, and other operating costs.

3.3 Test Pro g ram

Many things can influence the effectiveness of a corrosion inhibitor, and TVA has no previous experience with corrosion inhibitors in carbon steel raw water systems. It was decided therefore to conduct a test of inhibitor effectiveness under conditions which were as close as possible to expected plant operating conditions. Arrangements were made to conduct the test at Sequoyah Nuclear Plant in order to make the results directly applicable to the Sequoyah essential raw cooling water system.

The test stand included three parallel, I-inch, carbon steel lines each carrying approximately 15 gpm. Zinc polyphosphate was injected into one line at approximately 1.5 ppm. Sodium hypochlorite was injected into the second line at a rate intended to yield a chlorine residual of 0.6 to 0.8 ppm at the outlet of the line. The third line served as a control with no treatment of the raw water. A pipe sample was removed from each of the 3 lines on 4 occasions over a period of about one year. The volume reduction of the interior of each sample was measured to determine the extent of buildup.

3. 4 Results

Samples were removed after I, 3, 6-1/2, and 11-1/2 months of service.

Results of laboratory measurements on the samples are reported in References 22, 23, and 24. They have been plotted in Figure 4. The samples taken at 1 month and at 3 months showed rapid accumulation of deposit. Later samples showed a slower accumulation and in some cases an apparent reduction in deposits. The fact that some samples (par ticularly in the untreated line) indicated a reduction in deposits as time elapsed is probably because of random variations between samples.

All of the test samples had numerous small tubercles which were similar in appearance to one another and to the large tubercles observed in the sampling program described in Part A. The earlier samples had numerous tubercles up to about 1/16-inch high, but also had sizeable areas which were completely covered with tubercles up to about 1/8-inch high. The average diameter reduction in the later samples was on the order of 1/16-inch in the untreated line and the corrosion inhibited line and on the order of 1/10-inch in the chlorinated line.

In addition to random scatter which would be expected even under ideal conditions, operating problems contributed to imprecise results. Chemical injection and/or water flow were interrupted on several occasions because of mechanical problems with the feed pumps, empty feed tanks, and inadver tent power cutoff. Although similar problems are likely to occur in an operating plant system, the frequency of their occurrence here (particularly during initial operation) may have made test results non-representative.

Because of the variables discussed above, only three identified trends are considered valid. First, the rate of buildup was initially very rapid but had slowed dramatically by the end of the test. Second, the chlorinated line had more rapid buildup of deposits than the other two lines. Third, the zinc polyphosphate had little if any beneficial effect under the test conditions.

3.5 Conclusions and Recommendations

Neither zinc polyphosphate nor chlorination can be recommended for the control of corrosion on the basis of this test. It is possible that higher feed rates of zinc polyphosphate could be effective, especially during startup, and additional testing at higher feed rates may be desirable. It is recommended that further testing of zinc polyphosphate be carried out to determine whether a higher initial feed rate and/or more reliable injection would make it effective.

The tendency for chlorination to accelerate the buildup of corrosion products raises the question of whether current chlorination procedures should be altered to minimize this effect. The limited data provided by this test is not an adequate basis for answering this question. It is therefore reconanended that additional testing of chlorination be done to provide an adequate basis for an answer.

The observed high initial rate of buildup and lower rate later in the test are consistent with the sample data obtained in Part A and plotted in Figure 9. These data suggest that the rate of buildup may continue to decrease over a period of many years. If valid, this trend would indicate that there is little additional buildup after (for example) 25 years. Our existing samples would thus be adequate for 40-year pro jections even though no samples other than Watts Bar are older than 25 years. Additional testing should be done to verify this trend over a period of a few years.

It appears that corrosion resistant materials will be needed in many instances instead of carbon steel because of the serious corrosion problems with carbon steel and the lack of effectiveness of corrosion inhibitors. However, there is sometimes doubt as to the ability of a

supposedly corrosion resistant material to withstand a particular environment (stainless steel with low water velocity, for example).

The test stand can be modified to test some of these cases, and in fact is being modified for testing of stainless steel and a stainless steel/carbon steel weld joint. It is recommended that the stand be used for additional testing of various materials as the need arises.

REFERENCES

1. J.E. McKelvey note to Mechanical Engineering Branch Files dated November 5, 1976 (MEB761108012), "Browns Ferry Nuclear Plant - Emergency Equipment Cooling Water System Investigation of the Scaling of the Pipe Interior."

2. Memorandum from Roy H. Dunham to H. S. Fox dated February 23, 1977 (MEB770223076), "Scaling and Corrosion in Power Plant Raw Water Piping - GS-74."

3. Memorandum from Roy H. Dunham to H. S. Fox dated December 28, 1977 (MEB771228055), "Corrosion and Scale Accumulation in Raw Water Piping Systems at TVA Steam Plants - Pipe Sampling and Pressure Drop Testing -

GS-74."

4. Calculation PSS-OO-D053/0-HCG-WSB-020579-RO - "Average Reduction in Diameter of Raw Water Piping Due to Corrosion Product Buildup."

5. Calculation PSS-OO-D053/0-HCG-WSB-050579 - "Average Wall Reduction in Carbon Steel Raw Water Piping Due to Corrosion."

6. Cakulation PSS-OO-D053/0-HCG-WSB-050679 - "Maximum Wall Thinning in Carbon Steel Raw Water Piping Due to Corrosion."

7. Corrosion Study of Twelve Raw Cooling Water Pipes From Steam and Nuclear Power Plants - Power Service Center Lab Report No. C64-77-6105-3.

7A. Additional Raw Cooling Water Pipe Corrosion Data to Include Browns Ferry Nuclear Plant - Power Service Center Lab Report No. 64-77-6105-4.

8. Corrosion Study of Raw Water Supply Pipes - 15 Samples From Widows Creek Steam Plant - Power Service Center Lab Report No. 64-78-7141.

9. Corrosion Study of Raw Water Supply Pipes From Cumberland, John Sevier, and Watts Bar Steam Plants - Power Service Center Lab Report No.

64-78-7141-2.

IO. Corrosion Study of Three-Inch Raw Water Supply Pipe - Widows Creek Steam Plant - Power Service Center Lab Report No. 64-78-7141-3A.

11. Corrosion and Report Examination on Raw Water Supply Pipe - Kingston Steam Plant - Power Service Center Lab Report No. 64-78-7141-4.

12. Corrosion and Deposit Examination on Raw Water Supply Pipe - Gallatin Steam Plant - Power Service Center Lab Report No. 64-78-7141-S.

13. R. 0. Johnson note to Mechanical Engineering Branch Files dated November 3, 1977 (MEB771103003), "Long Term Effects of Corrosion and Scale Accumulation in Raw Water Piping Systems at TVA Steam and Nuclear Power Plants - Interim Report on Pipe Sampling Program."

13A. Memorandum from Roy H. Dunham to Harry G. Moore dated December 22, 1977 (MEB771222018), "Proposed Phipps Bend Nuclear Plant - Potential for Corrosion-Scaling in the Cooling Systems - N7M-KW."

14. Memorandum from H. G. Moore, Jr., to Roy H. Dunham dated August 29, 1978 (DES780830018), "Phipps Bend Nuclear Plant - Potential for Corrosion Scaling in the Cooling Systems - N7M-KW."

15. Phone conversations between R. E. Taylor, EN DES; Mike Tucker, Knoxville Utilities Board; and Dan Eckel, Ingersol-Rand.

16. Attachment to 45D from J. A. Raulston to D.R. Patterson dated Karch 2, 1979, "Corrosion on Pipe Exterior. "

17. Calculation PSS-OO-D053/0-HCG-WSB-022779 - "Summary and Correlation of Supporting Calculations for Corrosion Study of Pressure Drop Tests."

17A. Calculation of PSS-00-D053/0-HCG-WSB-020979 - "Analysis of Pressure Drop Data Taken in Carbon Steel Raw Water Piping Using the Darcy and Hazen-Williams Equations."

18. G. V. Shaw and A. W. Loomis, ed., Cameron Hy draulic Data, Fourteenth Ed.,

Ingersol-Rand Co., Cameron Pump Division, Woodcliff Lake, NJ 07675, pg. 27.

19. Crane Technical Paper No. 410 - Flow of Fluids Through Valves, Fittings, and Pipe, 1969, Crane Co., New York, NY, Fourteenth Printing, 1974.

20. Moody, Lewis F., "Friction Factors for Pipe Flow," Transactions of the ASHE, November 1944, pg. 671-684.

21. Design Guide DG-M3.5 - Pressure Drop Calculations for Raw Cooling Water Piping and Fittings.

22. Supplementary Corrosion Data on Raw Cooling Water Pipes from Steam and Nuclear Power Plants - Power Service Center Lab Report No. 64-77-6105-5.

23. Additional Corrosion Data on Raw Cooling Water Pipes - Power Service Center Lab Report No. 64-78-7309.

24. Additional Corrosion Data on Raw Water Pipes from Sequoyah Nuclear Plant Test Stand - Power Service Center Lab Report No. 64-79-7886.

7 7 7 7 7 7 7 7 7 7 7A Available Reference

Not Use Deposit

- on or Deposit Over-Wall Over-Wall Only Analysis Mn NA Brittle Brittle Layered -Large of Brittle Brittle Brittle Brittle Hard Brittle Red Brown Pipe Pipe j Hard Comments AeEearance Large, Tubercles Large, Tubercles Black, Deposit Amount Thin, Large, Tubercles. sized Large, Tubercles Large, Tubercles. sized Large, Tubercles Small Tubercles Small, Tubercles Large, Tubercles. Volume Performed 2

isis Mn0 NA NA NA NA NA NA NA NA NA s 0.3 0.8 0.1 72.4 0.5 2.2 3.3 1.0 2.3 0.6 0.5 0.4 NA Anal 2 Si0 NA NA NA NA NA NA NA NA NA NA NA

£OSit De 0 3 2

% Fe 82.9 88.I 3.7 75.8 81.4 84.4 13.4 38.6 90.4 87.5 NA In

. Wall g,

Max 0.073 0.077 0.079 0.061 NA 0.100 NA 0.122 0.046 0.036 NA Thinnin

, In Average Red 0.0266 0.0224 0.0155 0.0197 NA 0.0347 NA 0.0296 0.0136 0.0102 NA

\\ I Wall Samples I, In 25 Table Red Phase Average 0.273 0.167 0.822 0.070 0.133 0.314 0.198 0.218 0.054 0.096 0.163 Dia

Age.2. 7 2.0 5.0 (Years) 22.2 17.8 21.5 17.7 18 34.1 24 23.8 12.1

Line Line Supply Supply.

From Pump Pump Sample Supply Hose Core Cool Feed Water Line Feed Water Line Line Line Taken Sluice Protection Protection Line Rm.

System Backjet -Fire Hopper Hopper. Construction Boiler Cooling Ash Supply Boiler Cooling Fire Fire Alternate to FP Supply Ash Supply Ash Supply Tmp Water Discharge Spray

Flow Condition Continuous Continuous Continuous Stagnant Stagnant Nearly Continuous Stagnant Continuous Contiuous Continuous Continuous

Nom. 3 3 3 3 3 3 4 3 2\\ 4 2\\

Dia~

1 1 Bar Creek Creek Creek Piping Ferry Site 3 Sevier 2 Sevier 1 C 6 3 8 2

Colbert Unit Gallatin Unit John Unit John Unit Kingston Unit Watts Unit Widows Unit Widows Unit Widows Unit Sequoyah Const. Browns Unit

ID Sample C-1 G-1 JS-1 JS-2 K-2 WB-2 WC-1 WC-2 WC-4 SNP-1 BFNP-1

) 8 8 8 8 8 8 8 Available Reference

-Not Use in Wall in Wall in Seen Lost On Light

NA on or of Craters of Craters of Very Tubercles Lost Pipe Lost Pipe Lost Seen Appearance Difference Thickness Corrosion Pipe. Of Amt. in Amt. in Amt. Craters Deposit Wall Tubercles, Restriction Buildup, 80 Vari-colored Comments earance Wall Shipment; Wall

~ Large Deposit Shipment; Seen Large Deposit Shipment; Seen Large Deposit Shipment; Deep In Some In Craters Pipe Rust/Mud 0.4" Heavy Heavy Visible In Exterior Sch. Buildup With Deposit

~sis 3.1 2.3 2.8 2.8 1.9 1.5 S Al20a 7 1.6 Anal 0.4 0.5 0.8 0.9 0. 0.4 0.4

Si02 13.0 8.0 15.0 18.0 7.0 7.0 5.0 Eosit De.8

% Fe20a 89.0 80.8 72.3 71.5 80.0 72.5 77 In Wall g,

Max. 0.065 0.060 0.129 0.045 0.052 0.095 0.046 Thinnin Creek, In

Widows Red Average 0.026 0.018 0.016 0.012 0.015 0.054 0.027

- II Wall

Table, In Samples Red II Average NA NA NA NA 0.226 0.230 0.258 27 Dia Phase Age 25 25 25 25 25 25 25-1/4 (Years)

Line From Pump Pump Sample Discharge Supply Supply Discharge Feed Feed Sluice Taken System Cooler Cooler Cooler Cooler Piping Piping H 2 H 2 H 2 H 2 Boiler 4B Boiler 4C Bottom

Flow Condition Continuous Continuous Continuous Continuous Continuous Continuous Continuous

Norn. 6 6 4 4 2 2 2\\

Dia~

Creek Creek Creek Creek Creek Creek Creek Site 4 4 4 4 4 4 2

Widows Unit Widows Unit Widows Unit Widows Unit Widows Unit Widows Unit Widows Unit

ID Sample WC-6 WC-7 WC-8 WC-9 WC-10 WC-11 WC-12

9 9 9 9 9 9 9 9 Reference Available.5" Use Of Tubercles Few

-Not on or Some.2-.25" 0.5" Level, A Interior some Tubercles Deposits, Height Line Some Level, 2 Height 2 Tubercles some

NA Scattered Removed Orange Tubercles In For Appearance, earance Corrosion, Few In Analyzed, Dry Buildup Ht., Si0 1" Si0 Corrosion Tubercles Tubercles Collllllents.2-.25",

~ Mild Very Reddish Widely Tubercles, 0.4" Not Sample From Mild Tubercles, In Bright High Orange Throughout, to High Mild Except 0.3-0.5" Orangeish Most Muddy Streaked; 0.5"

ysis Al203 1.8 1.5 NA 1.4 2.0 2.7 1.9 1.5 S 0.4 0.6 NA 0.5 0.6 0.4 0.4 0.3 Anal

Si02 8.0 8.0 NA 9.0 21.0 17.0 9.0 14.0 posit De

% Fe203 88.5 88.9 NA 85.2 74.0 81.2 89.0 78.0 In Wall g, Sevier Max. 0.014 0.064 NA 0.027 0.075 0.019 0.057 0.055 Thinnin John, In

and Red Average 0.020 O.Oll NA 0.014 0.021 0.025 0.014 0.023 Wall (Continued)

Cumberland II, In

- Red 29 Average 0.096 0.162 NA 0.183 0.110 0.080 0.106 0.150 Table Dia Samples II Age 5 5 5 5 5 5 5 5 (Years)

Phase Line Vac. Vac.

To To To Bin Water Water Sample From Oil Cooler Water Water Pumps Water Water Taken Water Protection Oil Sluice Holding System Service P. Hx

Pulverizer Cooler Raw Fire BFPT Cooling L. Supply Cooling Priming Cooling Pump Cooling Pyrite

Flow Line Condition Continuous Stagnant Dry Continuous Continuous Continuous Continuous Continuous

Norn. 2 2 4 4 6 2 4 6 Dia~

Site Cumberland Cumberland Cumberland Cumberland Cumberland Cumberland Cumberland Cumberland

ID Sample CU-1 CU-2 CU-3 CU-4 cu-s CU-6 CU-7 CU-8

G-2.

12 12 12 12 III

Reference Available Ht. Phase Use Of the Not on or In

- Large Large - Of to NA Of Deposit. Indicating Reducing Exterior 1-1/4" Amount Comments earance refers

~ fV~nor Pitting. Number 1 to Reddish Large Sulfur, Presence Sulfate Bacteria 1.5 1.6 I Tubercles 1.4 1.4 study lsis S Al203 1.9 1.8 2.5 2.2 this Anal 2 10.3 8.3 8.2 in Si0 9.9 G-2 Eosit De to

% Fe203 78.2 74.8 82.6