ML21323A102

A New Look at Design of Raw Water Piping Bowman TVA 1980

ML21323A102
Person / Time
Site:	Browns Ferry
Issue date:	08/31/1980
From:	Bain W, Bowman C Tennessee Valley Authority
To:	NRC/RES/DRA
Aird, Thomas - 301 415 2442
Shared Package
ML21323A100	List: ML21323A101 ML21323A102
References
Download: ML21323A102 (5)

Text

, "

A new look at design of raw water pip.i:ng Extensive tests indicate that presently recognized standard m~tt,,ods for calculating pressure drops in raw water piping do not allow for required flow over a 40-yr projected operating lifetime ofthe,pipi_ng.

Changes in the recommended design practices are sugg~sted By CHARLES F. BOWMAN and WILLIAMS. BAIN, Tennessee Valley Authority Figure 1. Sample of piping from Kingston Steam Plant shows formation of deposit During preoperational testing of the raw cooling water system at TVA's Browns Ferry nuclear plant during the summer of 1976, it was found that certain heat exchangers were not receiving their de-sign water flow rates. Sections of the car-bon steel piping which supply water to these components were removed and were found to have a buildup of material on the interior which impeded the water flow. An analysis of the samples indi-cated that iron as Fe20 3 was the major constituent of the deposit. The buildup of iron oxide was attributed to oxidation of the pipe interior by common corrosion of steel piping by aerated river water. A study was undertaken to determine the pervasiveness of this problem in the TV A system and to develop recommended Figure 2. Average reduction in pipe diameter as a function of years of service practices to mitigate its effects in the l~

_i design of future power plants.

- Sampling program Approximately 50 sections of carbon steel raw water piping were removed from nine different TV A steam plants.

Both normally stagnant and normally flowing piping systems as well as both vertical and horizontal runs of pipe were sampled.

In virtually every case, the primary mech-anism was found to be corrosion of the steel piping by aerated river water and redeposition of the corrosion products.

The problem was found to a significant degree at all plants that were sampled.

The result is random pitting in the pipe wall and the formation of a tubercle over each pit as shown in Figure 1.

A measured length of each piping sample was sealed at one end and filled with water. The volume of water contained in the sample was compared with the origi-nal volume as calculated from the nomi-nal dimensions of a new pipe. The per-cent volume occupied by the deposit rep-resents the average loss in pipe cross-sectional area and was related to an average decrease in pipe diameter. This 0.....,, average diameter reduction as a function V of years of service is shown on Figure 2.

In addition, the deposit in each sample was removed and analyzed for various constituents as presented in Table 1.

From the scatter of data it can be seen POWER ENGINEERING/ AUGUST 1980 for pipelines with normally flowing or stagnant water; : * *,

• Sample taken from a.nor:nallyitagnan;li~e 0.6 0.5 WC-22 **

~

~ WC-f6 *

..... ~ 0.4

qi E:

0.3 0

'o

.s

0.

0

0 SNP-1

0./

a 0

5 15 20 25 30 35 Service years

,,/.'

-~

1

5

~

~-...... ~_,*~

---~ :':. -.. ~

Figure 3. Pictures of three samples of pipe removed from the same pipeline indicate the large variations that occurred in deposits.

40 73

t RAW WATER PIPING r al>le 1. Analysis of samples taken from deposits in raw water piping.

Sample Norn.

Flow System Sample Age Average

• % Deposit Analysis Comments on ID Site Diam, In. Condition Taken From (Years) Diam Red, In. Fe2 0 3 Si02 s Mn02 Appearance or Use C-1 Colbert 3

Continuous Boiler feed pump 22.2 0.273 82.9 NA 0.3 NA Large, brittle Unit 1 cooling water tubercles supply

("1 G-1 Gallatin 3

Continuous Ash sluice 17.8 0.167 88.1 NA 0.8 NA Large, brittle

~---

Unit 3 supply line tubercles JS-2 John Sevier 3

Stagnant Fire protection 17.7 0.070 75.8 NA 0.5 2.2 Thin, brown _deposit Unit 1 line K-2 Kingston 3

Stagnant Fire protection 18.2 0.133 81.4 NA 3.3 NA Large, brittle Unit 1 line tubercles WB-2 Watts Bar 3

Nearly Alternate supply 34.1 0.314 84.4 NA 1.0 NA Large, brittle Unit C continuous to backiet tubercles WC-1 Widows Creek 4

Stagnant Fire hose 24.7 0.198 13.4 NA 2.3 NA Large, brittle Unit 6 supply line tubercles WC-2 Widows Creek 3

Continuous Ash hopper 23.8 0.218 38.6 NA 0.6 NA Large, brittle Unit 3 supply line tubercles WC-4 Widows Creek 21/2 Continuous Ash hopper 12.1 0.054 90.4 NA 0.5 NA Small, hard tubercles Unit 8 supply line SNP-1 Sequoyah 4

Continuous Construction 2.0 0.096 87.5 NA 0.4 NA Small, hard tubercles const. piping water line BFNP-1 Browns Ferry 21/2 Continuous Discharge core 5.0 0.163 NA NA NA NA Large, brittle Unit 2 spray room coder tubercles WC-10 Widows Creek 2

Continuous Boiler feed 25 0.226 80.0 7.0 0.7 1.6 Rust/mud appearance Unit4 pump piping 0.4 in. tubercles WC-11 Widows Creek 2

Continuous Boiler feed 25 0.230 72.5 7.0 0.4 1.9 Heavy buildup Unit 4 pump pipinQ WC-12 Widows Creek 21/2 Continuous Bottom sluice 251/4 0.258 77.8 5.0 0.4 1.5 Sch. 80 pipe.

Unit 2 line Light buildup of tubercles with varicolored deposit WC-13 Widows Creek 21/2 Continuous Bottom sluice 25 0.115 77.9 7.0 0.3 2.4 Sch. 80 pipe, very Unit 2 line small buildup, 0.2 in. tubercles WC-16 Widows Creek 4.

Continuous H2 cooler 17 0.405 90.0

.6.0 0.1 3.0 Large amt. of buildup; Unit 7 cooling water Some tuber. tubercles r -.

3/4-1 in. high t

)

CU-1 Cumberland 2

Continuous Pulverizer oil 5

0.096 88.5 8.0 0.4 1.8 Mild corrosion, very coolinQ water few tubercles

~-

:

CU-2 Cumberland 2

Stagnant Raw service 5

0.162 88.9 8.0 0.6 1.5 Reddish deposits, water

• widely scattered tubercles, some 0.4 in. in height CU-4 Cumberland 4

Continuous Oil cooler 5

0.183 85.2 9.0 0.5 1.4 Mild buildup of

• cooling water tubercles, 0.2-0.25 in.

in height, some tuber-des briQht oranqe CU-5 Cumberland 6

Continuous Sluice 5

0.110 74.0 21.0 0.6 2.0 High Si02 level, water supply orange tubercles :

throughout, 0.5 to 1 in. in height CU-6 Cumberland 2

Continuous Cooling water 5

0.080 81.2 17.0 0.4 2.7 High Si02 level, to vac. priming mild corrosion pumps except for a few 0.3-0.5 in. tubercles CU-7 Cumberland 4

Continuous Cooling water 5

0.106 89.0 9.0 0.4 1.9 Orangeish interior to vacuum pump most tubercles 0.2-0.25 in.. some 0.5 in.

CU-8 Cumberland 6

Continuous Cooling water 5

0.150 78.0 14.0 0.3 1.5 Muddy appearance, to pyrite hold-streaked; some ing bin 0.5 in. tubercles JS-3*

• John Sevier 2

Stagnant Fire protection 23 0.242 85.6 6.0 1.0 1.0 Flaky deposits, Unit 1 system

( 1.3)

• large tubercles, 0.4 in. in hei ht JS-4 John Sevier 2

Stagnant Fire protection 23 0.277 91.0 6.0 1.3 0.6 Flaky deposits, darker Unit 2 system (0.8)* than JS-3, some tubercles 0.6 in.

JS -5 John Sevier 3

Stagnant Fire protection 23 0.177 85.6 6.0 0.7 1.3 0.5 in. tubercles system (1.0)*

WB-3 Watts Bar 4

Continuous Makeup to slag 35**

0.148 93.7 4.0 0.3 1.0 Relatively even Unit C chamber buildup, some tubercles 3/4 in.

in height WB-4 Watts Bar 6

Continuous Supply to ash 35**

0.150 92.5 5.0 0.4 1.2 SeeWB-3 Unit C quencher nozzles 74 POWER ENGINEERING / AUGUST 1980

Sample Norn.

Flow System Sample Age Average

% "Deposit Analysis Comments on ID Site Diam, In. Condition Taken From (Years) Diam Red, In. Fe 2 O3 SiO 2 s MnO 2 Appearance or Use WB-5 Watts Bar 2

Continuous Cooling water to 35**

0.206 89.7 5.0 0.5 1.2 Large number of Unit C boiler feed pump small tubercles WB_-6 Watts Bar 4

Continuous Makeup to slag 33**

0.152 81.8 8.0 0.3 1.5 See WB-3 Unit D chamber WB-7 Watts Bar 6

Continuous Supply to ash 33**

0.311 90.0 7.0 0.8 1.3 See WB-3, more Unit D quencher nozzles pronounced tubercles than WB-3 WC-21 Widows Creek 3

Nearly Water supply to 13.75 0.259 65.0 20.1 3.7 amples removed Unit 8 continuous pulverizers from linP. whP.rP.

WC-22 Widows Creek 3

Nearly Water supply to 13.75 0.440 79.1 6.9 2.4 pressure drop tests Unit 8 continuous pulverizers were oerformed. All WC-23 Widows Creek 3

Nearly Water supply to 13.75 0.326 81.0 8.1 2.4 samples had a large Unit 8 continuous pulverizers amount of buildup.

WC-24 Widows Creek 3

Nearly Water supply to 13.75 0.594 56.3 27.4 5.2 High SiO 2 levels in Unit 8 continuous eulverizers WC-21 and WC-24 K-7 Kingston 6

Nearly 23 0.169 81.8 5.4 1.3 Unit 7 stagnant K-8 Kingston 6

Nearly

{ f;c, pm<~foa 23 0.098 79.4 4.8 1 1 { Oaly a =*II *=*II Unit 7 stagnant line with a deposit but very large K-9 Kingston 6

Nearly very small 23 0.101 80.2 3.4 (2 in. in height) continuous 0.8 tubercles randomly Unit 7 stagnant flow spaced. Vertical line.

K-10 Kingston 6

Nearly 23 0.163 90.3 4.2 0.9 Unit 7 sta nant G-2 Gallatin 8

Stagnant Fire protection 19.5 0.227 78.2 9.9 1.9 1.5 Units 3 & 4 header G-3 Gallatin 8

Stagnant See G-2 19.5 0.313 74.8 10.3 1.8 1.6 { '-9' aumbe,of Units 3 & 4 large tubercles -

G-4 Gallatin 8

Stagnant See G-2 19.5 0.280 82.6 8.3 2.5 1.4 1 to 11/4 in. in height.

Units 3 & 4 Reddish deposit.

G-5 Gallatin 8

Stagnant. See G-2 19.5 0.359 75.5 8.2 2.2 1.4 Units3 & 4 NA - Not Available.

• Values indicated in parentheses indicate amount of MnO 2 found in John Sevier $amples only.

. * *Watts Bar was removed from service for a number of years so that these ages may not be the effective age.

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 and in some cases large varia-tions can be seen from samples removed from a single pipeline, as indicated by the three views in Figure 3 of a pipe from Widows Creek steam plant. It was found that the average buildup in 8-in.-diam samples taken at the Gallatin steam plant and 6-in.-diam samples taken at the King-ston steam plant was approximately the same as the buildup in the 2-in., 3-in.,

and 4-in.-diam samples taken at the John Sevier and Widows Creek steam plants.

Buildup does not appear to be dependent on pipe diameter.

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. Most of the samples had a relatively uniform buildup ( very rough surface). However, some samples such as shown in Figure 1 had almost no aver-age buildup but had large, randomly spaced isolated tubercles. No significant differences were observed in the cor-roded condition of horizontal and vertical t1 *) runs of pipes as long as the pipes were

~

completely full of raw water.

Since TV A does not have any steam plants which have been in service for 40 years, and due to the large scatter in the POWER ENGINEERING/AUGUST 1980 observed data, an approach was devel-oped to estimate a 40-year design value of diameter reduction from the available data. 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 in., was selected as the 40-year design value. Although the. ap-proach 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.

Pressure drop tests Tests were performed at the Widows Creek, Kingston, and Gallatin steam plants to evaluate the effects of corrosion product buildup on pressure drop. 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.

An orifice inserted in a length of new pip-ing 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 measure-ments were taken. Taps were installed in the lines to allow pressure drop measure-ments to be made and mercury manome-ters were used to measure the pressure drops across the orifice and each section of piping. Samples removed from each test line were analyzed to determine the percent volume reduction of the pipe inte-rior due to the corrosion product buildup.

It was found that the samples removed from the 3-in. test line at Widows Creek had a substantial amount of buildup on the interior, made up of iron oxide and silicon oxide, whereas the samples re-moved from the 6-in. line at Kingston were found to have only a small amount of uniform buildup but had very large ran-domly spaced tubercles ( some ap-proaching 2 in. in height). The 8-in.-diam samples from Gallatin were found to have a more uniform buildup than the Kingston 6-in. line but also had large, randomly spaced tubercles. The average diameter reduction of each set of samples was found to be the following:

Plant Widows Creek Kingston Gallatin Section A Section B Average Measured

• Diam Reduction 0.405 in.

0.133 in.

0.270 in.

0.320 in.

75 *

~

I RAW WATER PIPING The corresponding diameter reduction for each test line was then used with the pressure drop test data to develop ap-propriate equations for predicting pres-sure drop. The Hazen-Williams and Dar-cy equations for pressure drop were con-sidered with each set of data being treated separately and then analyzed to establish a correlation to the other sets of data. Predictive equations were tt:ien for-mulated to predict pressure drop iri car-bon steel raw water piping after 40 years of service.

Correlation of test data The Hazen-Williams equation can be writ-ten in the form hl = 0.2083 ( 1~) 1*

85 Q1.65/d*.8655

( 1) where:

hL = head loss in feet per 100 ft of pipe d = pipe inside diameter in inches C = roughness factor Q = flow rate in gpm A least squares curve fit of the form hL = a1Q1.s5

( 2) where a1 is a constant, was obtained for each set of data. Equations ( 1) and (2) were set equal to solve for d:

d = [ (0.2083/a1) ( 1~) 1.85] 1/4.8655 (3) 18

~ 7 i5

~

~6---+----+-+-+-6-H--#1------i lg Cl)

~ 5 f---t----t-<e--f-:!~'-7">-r---;

• I ~4---+------r--,--~

~

Cl)

~31----t----<Jf-f-:trn#~-='-i 0

~

~2r---t---ffj~-+----t--~

0 :;

u ci u

40 60 80 JOO

• 76 For each set of data ( a unique value of a1), a table of ( C,d) values was gener-ated which will satisfy Equation ( 3).

Several figures were generated in an attempt to find a correlation between diameter reduction and C. Values of C were assumed and corresponding values of d were calculated for a given test using Equation ( 3). A dimensionless parame-:

ter, d 0

, was defined for use in correlating the above calculated value of d with the measured value of diameter reduction:

d. = (dNOM -

dCALC)

~dMEAS Calculated Diameter Reduction Measured Diameter Reduction (4) where:

dNOM =

dCAt.c=

nominal inside diameter of new pipe, in inches calculated inside diameter of pipe using Equation ( 3) diameter reduction corre-sponding to the percent vol-ume reduction measured, in inches d

• as a function of C for all the pressure drop tests is shown on Figure 4. The smallest variation of calculated diameter reduction I measured diameter reduction occurs at a value of C of approximately 57 at a value of d* equal to 2.

Using a slightly more conservative value of C equal to 55 and a diameter reduction equal to twice that measured, the Hazen-Williams equation becomes hl = (d 2 ~d

)4.6655 NOM MEAS 0.63Q1.85 (5)

The Darcy equation can be written ( for a pipe length of 100 ft) in the form hl = 3.11 fQ2/d5 (6) where hL, Q, and d are the same as defined previously and f is the friction factor. A least squares curve fit of the form Figure 4. ( left) Relationship between rough-ness factor and the ratio of calculated diame-ter reduction to measured diameter reduction

( for Hazen-Williams equation) is shown for all pressure drop tests.

Figure 5. ( right) Relationship between abso-lute roughness of pipe interior and ratio of cal-culated reduction to measured diameter re-duction (for Darcy equation) is shown for all pressure drop tests.

hl = 8202 (7) where a2 is a constant, was obtained for each set of data.

Setting Equations ( 6) and ( 7) equal and solving for the friction factor, f:

f = a2d5 (8) 3.11 Moody presented an expression for the friction factor in fully rough flow as:

~

3.7 1 / V' = 2 log1oic/d (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:

Eld=3.7/[101I211']

(10)

For each pressure drop test ( a unique value of a2) values off were calculated for different assumed inside diameters using Equation ( 8). Equation ( 1 O) was used to calculate Eld. Using the assumed values of d, the corresponding values of E 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 E vs d*, shown in Figure 5.

The smallest variation in E appears to occur at a value of d

• equal to 1.0 to 1.2 where E = 0.9 to 0.8 in.

Using a value of E = 0.9 in. and a calcu-lated diameter reduction equal to the measured value of diameter reduction 5.0

.s c_i0. 02.____.__.....__.____,_ _ _._~

0 0.5 /.0 1.5 2.0 2.5 3.0 d*, calculated diam. reduction/

measured diam. reduction POWER ENGINEERING / AUGUST 1980 0

0 Figure 6. ( right) Comparison of measured and predicted pressure drops for Widows Creek 3-in. line.

cu 40f:---........i""1l'cy.-iWll,~~~~'4---t----:-.d~d'-t---i Q.

Rgure 7. ( below) Comparison of measured and predicted pressure drops for Kingston plant 6-in. line.

~ 301----:...::='--'1:..:..'..'.!=~C!!!!--+-,,,.@~7"'-i,!+---I---I 8

Rgure 8. ( below right) Comparison of measured and predicted pres-sure drops for Gallatin plant 8-in. line.

20 l----+-----+--4-..,;L-~-+---+--t-----i

~ IO 1---+-:;;;;...-J-,,,,"-==--+---+--t---t---t-- -;

"tJ 0

Cl>

t Flowrate, gpm Measured data

, measured

-- Hazen-Williams with C =55 and diam. =nom.diam.-211D >

Measured data,Section I

--- Darcy with :=0.9in. and diam. =nom.diam. -11D measured Measured data, Section2

-*- Hazen-Williams withC=IOOanddiam. =nom. diam.

2 5 -- Hazen-Williams with C =55 and 12~-~-~-~-~-~--~~

Q.

0 0

0 ' -

......,,~

~

"tJ 0

t 10 8

6 4

2 C> t 20 l-r...-il---+--+,,--~---11---+r--+----i

~

~

51---1--t--+~IF..,,.l,m~--t--:--t----1 "tJ 0

150 200 250 300 350 400 450 Flowrote, gpm

£

~OOL--"==4CO=::........iL-..6CXJ........._____,__EXXJ_._____.__l_(XXJ....____.

Flowrate, gpm

( d * = 1 ), the Darcy equation becomes hl = 3.11 fQ2/(dNOM -

MMEAS) 5 (11) where f ={210910 [ ( 4.1) ( dNOM - t.dMEAS) l }-2 (12)

Predicting pressure drop 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.

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. Based on the results of the sampling program, this value is 0.4 in.

as indicated above.

Substituting these values into Equation

( 5) the Hazen-Williams equation to cal-culate pressure drop ( head loss per 100 ft of pipe) in piping after 40 years of service is:

0.63O185 h =

L (dNOM -

0.8)*-8655 (13)

Similarly, substituting the 0.4 in. value of diameter reduction into equations ( 11 )

and ( 12), the resultant Darcy equation to calculate pressure drop (head

• loss POWER ENGINEERING / AUGUST 1980 per 100 ft of pipe) in piping after 40 years of service is:

hl=3.11fQ2/(dNOM-0.4)5 (14) where:

f= [210910[(4.1) (dNOM-0.4)JJ-2 (15)

At reasonable velocities, the modified Hazen-Williams equation consistently predicts values of pressure drop greater than the modified Darcy equation when Equations ( 13) and ( 14) are used. This is apparently because the large diameter reduction projected after 40 years is dou-bled 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 modi-fied Hazen-Williams equation always pre-dicts higher values of pressure drop than the modified Darcy equation.

Note that as diameter increases, the val-ue of dd and f [from Equation ( 15) J decrease. Since both of these values decrease, the modified Darcy head loss prediction decreases more rapidly than would be seen based on diameter in-crease alone. This is in contrast to the constant C factor incorporated in the modified Hazen-Williams equation.

Therefore, for more conservative calcula-tions, the modified Hazen-Williams equa-tion should be used.

Impact on design The results of this study should have a profound impact on the design of future raw water piping systems, since the pres-sure drops calculated by methods rec-ommended herein are significantly great-er than those which would be calculated by presently recognized standard meth-ods. For example, for a new pipe design velocity of 4 fps, the pressure drop calcu-lated by the methods recommended herein in 100 ft of straight 3-in. and 8-in.

line would be approximately 1-2 and 4 times, respectively, that calculated by the Hazen-Williams formula with the com-monly accepted C value of 100. Alter-nately stated, a 3-in. line sized by con-ventional means to pass 95 gpm will be capable of passing only 25 gpm with the same available pressure drop, or an 8-in.

line sized to pass 1000 gpm will pass only approximately 450 gpm at the end of its 40-year life.

It should be noted that these results apply only to friction losses and should not be applied to form losses.

END 77