Cooney) WATER wy Technical Note A Simple Correction for Slug Tests in Small-Diameter Wells by James J. Butler Jr! Abstract ‘A simple procedure is presented for correcting hydraulic con ductivity (K) estimates obtained from slug tests performed in small-diameter installations sereened in highly peemeable aquifers, Previously reported discrepancies between results from slug tests in small-diameter installations and those from tests in nearby Jarger-diameter wells are primatily @ product of frictional losses within the small-diameter pipe, ‘These frictional losses are readily incorporated int existing models for slug tests in high-K aquifers, which then serve asthe basis ofa straightforward procedure for cor recting previously obtained K estimates. A demonstration ofthe peo- posed procedure using data from a series of slug tests performed in ‘controlled field setting confirms the validity ofthe approach. The results of this demonstration also reveal the detailed view of spa- tial variations in K that can be obtained using slug tests in small diameter installations. Introduction ‘Slug tests in small-diameter wells or direct-push installations cean be @ useful tool for obtaining information about variations in hydraulic conductivity (K) ata scale of relevance for contaminant transport investigations (Hinsby etal. 1992; Yeh e al. 1995; Scholl and Christenson 1998; Butler etal. 2002). Well diameter, however, cam influence tests in imervals of very high hydraulic conductivity Figure 1 compares K estimates from a series of slug tests per- formed in small-diameter direct-push installations (0.016 m inner diameter [LD with those from multilevel slug tests in nearby mon- itoring wells (0.102 m LD.) sereened in the same coarse sand and gravel aquifer, Inthe lower, more permeable portions ofthe aquifer, the direct push estimates are much smaller than those from the con: ventional monitoring wells. Butler etal. (2002) have shown tha the small diameter ofthe direct-push rods isthe primary factor respon: "Kansas Geological Survey, 1930 Constant Ave., Campus West, University of Kansas, Lawrence, KS_ 66087; (788) 864-2116; jbutler@ kgs.ukans edu Received October 2001, accepted January 2002, Vol. 40, No. 3 GROUND WATER— May-June 2002 Comparison of Multilevel Slug Tests with Direct-Push Slug Tests (0.016 m ID Rods) cy b eo Gemean [FS eeome EES Depth Below Datum rr ee er a) Hydraulic Conductivity (miday) Figure 1. Comparison of K estimates from direet-push (DP) slug tests ‘with estimates from multilevel slg tests performed at wells Gems4S ‘and Gems4N (length of test interval for multilevel sug tes darkened symbols designate tests near Gems4N; open triangles des- Jgnate tests near Gems4S; lateral separation between Gems4N and (Gems4S is 9.7 ms A is test interval for Figure 2s after Butler etal, 12002). sible for this underprediction in K. Given the significant logistical and budgetary advantages of sinal-diameter installations, a method for correcting slug-test K estimates for well diameter would be of considerable value, The description and field verification of such ‘a method is the major purpose of this note. ‘One would expect that as the well diameter decreases, head losses due to friction within a well would increase, If reasonable K estimates are to be obtained from small-diameter wells, the model sed for analysis of the sug-test response ta must incorporate fric- ‘ional well losses, Van der Kamp (1976) developed relationships for assessing when frictional well losses can be ignored in laminar and turbulent flow regimes, but did not consider the ease when such losses cannot be neglected. Ross (1985) incorporated frictional 303‘well losses into the model of Van der Kamp as part ofa study of slug tests in wells with very long water columns, but did not discuss tests in small-diameter wells, Butler (1998) showed how the frictional loss term of Ross can be incorporated into the high-K extensions of the Bouwer and Rice and Horsley models, but also did not con- sider the utility ofthe approach for small-diameter wells. Apparently, no previous work has specifically addressed the issue of frictional ‘well losses during slug tests in small-diameter well, In this note, the approach of Van der Kamp and Ross is used as the basis of a method for correction of K estimates from slug tests in small-diameter wells, brief overview ofthe theory is frst pre- sented, after which the correction approach is described. The appro- priateness of the method is then assessed using results from a series of slug tests performed in a controlled field setting, The presentation ‘concludes with a summary of the major findings ofthe study. Overview of Theory and Correction Procedure Ross (1985), following on the work of Van der Kamp (1976), presented a theoretical derivation ofa slug test model that includes, frictional losses in the well. For the laminar flow conditions, assumed in that derivation, the frictional fore per unit length ofthe well easing (F) can be defined as w where p = density of water 1 = kinematic viscosity of water \w = deviation of water level in well from static conditions, ‘Equation 1 can be incorporated into a momentum balance for the wel. If nonlinear terms are neglected and flow between the well and the aquifer is represented by a quasi-steady-state flow model (Hvorslev 1951; Bouwer and Rice 1976), the following equation for slug tes in a high-conductivity aquifer can be obtained: uw , (BL, Se) aw + (Mee oe Leg ( 2 ae) arene oO were sete length wie) ‘Zarbuchen et al. 2002) Pro Myo ae TO get MenO @ 304 where G oF dimensionless damping parameter = fl 88, fe BVel, * 3K, VL, {4 = dimensionless time parameier ((p/L,)!*) w= normalized water-level deviation (w/H,) = change in water level initiating a slug test (intial displacemen), Equation 3 is in the same form as the equations for the high- K extensions ofthe Bouwer and Rice (Springer and Gelhar 1991) and Hivorslev (Butler 1998) models. The only difference is that the Co parameter has been redefined to incorporate frictional losses in the wel. The analysis process using Equation 3 is the same as that described in Butler and Garnett (2000) fr the case in which fric- tional well losses are neglected, The same set of Cp type curves is, superimposed on a normalized plot of the response data, and hydraulic conductivity is calculated by substituting values for the \well-construction parameters, the damping parameter fr the best fit type curve (Cp*), and the match points from the actual ((*) and dimensionless (1) time axes into the appropriate equation: Unconfined: High-K Bouwer and Rice Model a in eu) “2(o- ee) a Confined: High-K Hvorslev Model k= Fen [b/(2r,) + (1 + (b/(2,))°)4 e » SVL ty ©) as{cs-BE 8) where R,= effective radius parameter of Bouvier and Rice (1976) 1,,= radius of well sereen ot borehole in isotropic aquifers. For anisotropic aquifers, 1, should be replaced by 1, VK,/Ky. where K, is vertical hydraulic conductivity Zlotnik 1994) ces eV ‘The subtraction in the denominator of Equations 4 and $ rep- resents the correction for well diameter. The comection term has been rewriten in terms of for ease of implementation in the analy sis process. The importance of the correction depends on the radii of the well easing and screen, and the Tength of the water column, in the well. Clearly, as the radius of the well casing increases and/or the length ofthe water colurmn decreases, the correction term ‘becomes less significant. The correction is likely o be unnecessary for wells greater than 0.012 m in radius. As can be seen from Equations 4 and 5, a small C’, value and a long column of water ‘would be required for the correction tobe of significance in that sit- uation, Note that it is not necessary for previously analyzed slug tests tobe re-analyzed to account for well diameter. The original K, esti-DP43C - 0.61 m screen 4/5/00 Slug Tests en Ww Test 1-004 Test 2-H-0.185m + Test 8-Hj=0.246 m Test 6 - H,=0.082 m = Type Curve - C5°=0.346 . 5 0 % oo 2 w% Time (sec) Figure 2. Normalized head (w(t)/Fl, where w(t is deviation from sta- tic and Hy, is magnitude of initial displacement) versus time plots for April 5,200, sug tests at DPA3C (0.016 m LD.) with the best-fit type ‘curve from the high-K Hvorsley model. mate can be corrected for well diameter through multiplication by the ratio of original over corrected Cy, estimates: u am a © Field Verification The correction for well diameter presented in the previous section is based on the assumption tha frictional well losses can be represented by the linear loss term characteristic of laminar flow con- DP-0.23m seroon + DP -0.50m sorean DP -0.61 m sorean Depth Below Datum (mn) tor P4236 and 8 i wD —s pane - “Se! . ? 2 a a a ee Hydraulic Conductivity (m/day) Figure 3. Comparison of corrected K estimates from direct-push (DP) slug tests with estimates from multilevel slug tess performed at wells Gems48 and Gems4N (same conditions as in Figure 1; K correction consisted of multiplication by Equation 6; arrow labels defined in text. the same wells, so they were considered as standards to which the corrected K values from the small-liameter drect-push installations ccould be compared. Figure 3 displays the multilevel slug tests from the two monitoring wells and the direct-push K estimates cor- rected by multiplying the original values by Equation 6. The agree- ‘ment between estimates from the direct-push and multilevel slug tests is much better than in Figure 1. At the depths in the vicinity of the arrow labeled DP43B and DP43C, two sets of direct-push installations were located 0.84 m apart and screened over the same interval. At both installations, the screen length was progressively increased between tests as described in Butler et al. (2002), Because the results from the tests atthe two installations were quite similar, ‘only K estimates from the three sets of tests at DPA3C are presented ‘on Figure 3. Similarly, at the depths in the vicinity of the arrow labeled DP464 and DP424E, slug tests were performed at two installations located 1.26 m apart and screened over the same inter- val. Because the estimates were again quite similar, only the K value from DP464 is presented. Note that the estimates from tests per- formed inthe installations with the shortest screen (0.08 m) differed significantly from the results of the multilevel slug tests. Ths dif ference is likely a product of small-scale aquifer heterogeneity However, the lower 0.08 m installation was emplaced in a manner that often leads to significant screen blockage (Butler et al, 2002), so the relatively low K value at that Ievel could also be a product of partial sereen blockage. ‘The results from tests at DP43B and DP43C can be examined in more detail to further assess the viability of the correction pro- cedure, Figure 4a is an expanded view of the relevant depth inter- val with the K estimates from the nearby monitoring well (Gems4N = 1,85 m from DP43B) and the original K estimates from the (wo direct-push installations. At both installations, the original K esti- ‘mates for the test interval decreased with respect to K values at 305[= Seneait- 067 miners Page 015 macreen| 160 Orato. 0st macean = Beate -0 15 mecrer - SE Bates 020-m aceon E ies == dese 081 maceon 2 16. Si a7 av é 174 176 178 180 0% 7S CO Hydraulic Conductivity (m/day) rected K estimates from DP43B and DP43C (DP43B-Gems4N lateral separation = 1.85 n DP43B-DP43C lateral separation = 0.84 m). 160 162 | 164 166 168 170 wa Depth Below Datum (m) ia 17.6 | [= Goneiv-0.51 morale = DPA yal [5 pease 180 0 SDSCSSCHOOSCTBSCSD Hydraulic Conductivity (m/day) DPAIC-Gems4N lateral separation = 2.16 m; 16.79-1740m 22 582 ‘Table 1 Original and Corrected K Values for Shug Tests at Direet-Push Installations DP43B and DP43C Test Original Corrected Correction L & Interval K (mmfday) K (ividay) Factor® tm) C Pan 17.16-1731 m ona 1019 1st 1096 093 1701-1731 m 532, 95.1 179 10.78 0:98 16.74-17.35 m 390 7 1.98 1062 096 pase 1723-1738 m 649 98.1 131 11.16 093 17.10-17.40 m 603 M161 192 1092 098 2.09 10.56 0.96 inert vison (= 120% 10-6 mh (war empert 132°C ctv casing rai, = 0.0074, serra) = 0.0127 m, Gems4N as the length of the interval increased, as would be expected in the presence of significant well losses The significance of the well losses is demonstrated in Table 1, where the original K, estimates are presented along with the corrected estimates obtained through multiplication by Equation 6. ‘The corrected estimates in Table I can also be used to obtain « detailed view of spatial variations in K in the vicinity of these installations IF the aquifer is assumed to be perfectly layered in the region affected by the slug fests, then the K estimate forthe test inter- val can be assumed tobe the arithmetic average ofthe layer values Furthermore, if the new material exposed as the test interval is increased in length is considered a layer, then layer K values can be calculated using the corrected estimates of Table 1, The resulting layer values are shown in Figure 4b and allow mote insight to be obtained regarding lateral and vertical variations in hydraulic con- ductivity in the vicinity of DP43B, DP43C, and Gems4N. For 306 ‘example, a significant change in K was observed at Gems4N in the Vicinity of 17.3 m, while tis change occurred at approximately 17.0 (o 17.1 mat DP43B and DP43C. Although K estimates forthe upper and lower layers at DP43C are in good agreement with those from simiar intervals at DP43B and Gems4N, the high-K layer found at DP43C from 17.1 to 17.25 m was not observed atthe other instal lations. The level of detail ilustrated in Figure 4b clearly demon- strate the potential of direct-push methodology for detailed descrip- tions of spatial variations in hydraulic conductivity. Such details, however, would not be recognized without the correction procedure described here, Summary and Conclusions Previous work hus showin that slug tests in small-diameter installations in aquifers of high hydraulic conductivity will yield K estimates that can be considerably lower than those obtained atnearby conventional wells (0.05 m LD. of large) seeened in the same aquifer, The work desried in his note demonstrated that frie- tional well losses are the primary factor responsibe for this under- prediction in K. A simple procedure was proposed to correct hhydraulic conductivity estimates for well diameter. This proce- ‘dure, which can be readily incorporated into existing methods for analysis of slug tests inhigh-K aquifers, was verified using a series of tests performed in a controled field setting. The results of this field assessment also demonstrated the potential of slug tests in small-diameter installations fr detailed description of spatial vai- ations in hydraulic conductivity, ‘The correction procedure described in this note assumes that well losses can be represented using a linear laminar-flow model Although a linear model may not always be valid its appropriate” tess can be assessed for any particular field setting using normal- ized plots of slug-test dat, If the normalized plots coincide for tests inated with diferent inital displacements, suchas in the case ofthe tests desribed her near model of well loses i appropriate Acknowledgments This research was supported in part by external funding pro- vided by the Kansas Water Resources Institute under grant HQ96GRO02671 Modif. 008 (subaward $0104). Field assistance provided by John Healey of the Kansas Geological Survey, and Elizabeth Garnett and Wes MeCall of Geoprobe Systems is rate- fully acknowledged, References Bouwer, H., and R.C. Rice. 1976, A slug test for determining hydraulic conductivity of unconfined aquifers with completely or partially penetrating well. Water Resources Research 12, no. 3 423-428, Butler, 1, Jr 1998. The Design, Performance, and Analysis of Slug Tests. Boca Raton, Florida: Lewis Publishers. Butler JJ. Jr, and EJ. Garett 2000, Simple procedures for analysis of sg tests in formations of high hydraulic conductivity using spread- sheet and scientific graphics software. Kansas Geological Survey Open-File Rept. 2000-40 (available at www.kgs.ukans.edu/ Hydro/Publications/OFROO_AOindex. ral), Jr, 1M. Healey, G.W. McCall, EJ. Gamet, and S.P.Loheide 1, 3002. Hydraulic tests with ditect-push equipment. Ground Wazer 40, no, 1: 25-36. Hinsby, K., PL. Bjerg, LJ. Andersen, B. Skov, and E.V. Clausen, 1992, ‘slug test method for determination of a local hydraulic ivity of an unconfined sandy aquifer. Journal of Hydrology 136, 87-106, Horsley, MJ. 1951. Time lag and soil permeability in ground-water ‘observations. Bullen no. 36. Waterways Experiment Station, Corps of Engineers, U.S. Amy. Kipp, KL, Jr 1985. Type curve analysis of inertial effets in the response ‘of a well to a slug test, Water Resources Research 21, no. 9 1397-1408, Ross, B. 1985. Theory ofthe oscillating sug test in deep wells. Memoirs ‘of temational Association of Hydrogeologiss V7.no. 1:44. 17d International Congress on the Hydrogeology of Rocks of Low Permeability Scholl, M.A. and S. lhristenson. 1998, Spatial variations in hydraulic con- ductivity determined by slug tests inthe Canadian River alluvium neat the Norman Landi, Norman, Oklahoma. U'S. Geological Survey ‘Water Resources Investigations Report 97-4292. Springer, RK., and L.W. Gelhar. 1991. Characterization of large-scale aquifer eierogencity in glacial outwash by analysis of slug test with ‘oscillatory response, Cape Cod, Massachusetts. U.S. Geological Survey Water-Resoutees Investigations Report 91-4034, pp. 36-40. ‘an der Kamp, G. 1976, Determining aquifer transmissivity by means of ‘well response tests: The underdamped case. Water Resources Research 12, 90,1: 1-77 Yeh, T-C. J, J. Mas-Pla, TM. Williams, and J.P. McCarthy. 1995. ‘Observation and thre dimensional simulation of chloride plumes in sandy aquifer under forced-gradient conditions. Water Resources Research 31, no, 9: 2141-2157, Zlotnik,V.A. 1994. Interpretation of slug and packer tests in anisotropic aguiless. Ground Water 32, no, 5: 761-766, Zurbuchen, B.R., VA. Zlotnik, and JJ. Buller Jr. 2002. Dynamic inter- ‘pretation of slug ess in highly permeable aquifers. Water Resources Research, in ress. 307