PETROLEUM SOCIETY

CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM PAPER 2007-191

Determination of Real-Time Dynamic Fluid

Levels by Analysis of the Dynamometer Card
Z. LIU, H. WANG
China University of Petroleum

D. YANG
Petroleum Technology Research Centre (PTRC), University of Regina

This paper is to be presented at the Petroleum Society’s 8th Canadian International Petroleum Conference (58th Annual Technical
Meeting), Calgary, Alberta, Canada, June 12 – 14, 2007. Discussion of this paper is invited and may be presented at the meeting if
filed in writing with the technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will
be considered for publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to
correction.

Abstract and annular gas. In addition to determination of the real-time

dynamic fluid level, this method greatly reduces the production
Although the beam pumping is the dominant pumping system maintenance cost. A PC-based software for determining the
in the onshore oilfields, many issues pertaining to its hydraulic dynamic liquid levels is also developed.
performance have not been well understood. This is mainly
ascribed to the fact that it is difficult to determine the downhole
pump performance data, such as the dynamic fluid levels. In Introduction
practice, the dynamic fluid levels are traditionally measured on-
site by using the acoustic method. This method, however, finds The dynamic fluid level, defined as the distance from the
its limitations in determining the real-time dynamic fluid level. wellhead to the fluid level in the annulus, is a key pumping
Recently, the automation technology has been widely used in parameter to be measured on a regular basis. It has been well
the oilfields, it is essential to measure the real-time dynamic accepted that fluid level survey is considered as the primary
liquid levels for optimum reservoir management. In this paper, diagnostic tool for monitoring downhole status and confirming
a method is developed to determine the real-time dynamic liquid the problem[1]. In general, the dynamic fluid level can be used to
levels by analyzing the measured dynamometer card. determine the production strategies, analyze the reservoir
Theoretically, a mathematical model is developed to determine performance, and to estimate the reservoir pressure at the
dynamic fluid levels by using the average polished-rod loads various stages of field development since it is constrained by
(PRLs) for the two turning points at the end of the upstroke and the formation pressure and the production rate[2, 3]. In addition,
downstroke, respectively. In practice, the average PRLs can be the dynamic fluid level together with other pumping parameters
directly obtained from the dynamometer cards and then used to can be used not only to determine the downhole pumping status,
determine the dynamic fluid level. It has been found from field but also to perform well diagnosis and maintenance[4].
applications that this newly developed method provides
At present, numerous techniques have been developed to
accurate determination of the dynamic fluid levels which will
measure the dynamic fluid level, among which the acoustic
not be affected by the existence of noise, dogleg, tubing collars,

1
method is the simple and widely used one. More specifically, an traveling valve is open, while the standing valve is still closed.
acoustic wave is first generated at the surface, transmitted to the Obviously, the PRL is mainly composed of the weight of the
dynamic fluid level in the annulus, and then reflected back to rods suspended in the fluid, the tubing head pressure (THP), and
the surface. The time interval is recorded from the emission of the friction between the plunger and the barrel. If the loads due
the acoustic wave to the return of its reflection to the surface. to acceleration and vibration are neglected, the PRL at point a
Then the dynamic fluid level in the annulus is determined as the can be calculated by
velocity of the acoustic wave in the air is known[5]. In practice, PFa = − Fplunger + Wr − W f − Phd (1)
it is difficult to use this technique for measuring the real-time Wr = ρr gLAr (2a)
dynamic fluid levels where the automatic monitoring system is
W f = ρ l gLAr (2b)
installed due mainly to its limitation of continuously generating
the acoustic wave[6]. Air gun can be used to generate continuous ρl = f w ρ w + (1 − f w )ρo (2c)
high-energy wave; however, the cost is high and its life Phd = Ar poil (2d)
expectancy is short[7]. In addition to the acoustic method, the
dynamic fluid level can also be obtained by measuring the where PFa is the PRL at point a, N; Fplunger is the friction
relationship between the casing pressure and time when the between the plunger and the barrel, N; Wr is the weight of the
casing gas is vented into the air[8]. However, this approach is
limited by the gas volume in the annulus. rods in the air, N; W f is the buoyancy force on the rods, N;
The surface dynamometer card represents the polished-rod Phd is the force imposed by the THP at the end of downstroke,
load (PRL) versus its position, which is usually measured by
using a dynamometer[9-15]. Proper analysis of the dynamometer N; L is the length of the sucker rod string, m; Ap and Ar are
usually can be used to reveal the following information[14]: 1) the cross-sectional area of the plunger and the sucker-rod,
whether the downhole pump valves are leaking and the extent respectively, m2; g is the gravitational acceleration, m/s2; poil
of leaking; 2) overall performance of the downhole pump; 3)
minimum and peak loads and stresses; 4) the PRL at each point is the THP, Pa; f w is the water-oil ratio; and ρ l , ρo , ρ w and
in the stroke cycle; 5) the torque applied to the gear reducer and ρr are densities of fluid, oil, water, and steel rod, kg/m3,
prime mover; 6) the polished-rod horsepower; and 7) the respectively.
counterbalance effect at the polished load. Recently, with At point b, the standing valve is still closed as shown in
development of the automatic monitoring system in the
oilfields, the surface dynamometer card can be transmitted to Figure 2 because the intake pressure Pi is lower than the
the workstation by a transceiver or the global positioning pressure P2 in the chamber of the plunger. At this moment, the
systems (GPS)[16]. So far, no attempts have been made to
determine the real-time dynamic fluid level by analysis of the pressure, Ps , above the traveling valve and the pressure, P2 , in
dynamometer card. the chamber of the plunger are considered to be the same,
In this paper, efforts have been made to determine the real- though the traveling valve is just closed due to its gravity. This
time fluid level by analyzing the measured dynamometer card. is why the PRL does not show a significant increase and the
Theoretically, a mathematical model is developed to determine stroke remains zero in the measured dynamometer card. Thus,
dynamic fluid level by using the average polished-rod loads the PRL at point b also consists of the weight of rods suspended
(PRLs) for the two turning points at the end of the upstroke and in the fluid, THP and opposite friction between the plunger and
downstroke, respectively. In practice, the average PRLs can be the barrel. If the loads due to acceleration and vibration are also
directly obtained from the measured dynamometer cards since ignored, then the PRL at point b, PFb , can be described as
the determined PRLs are found to be in good agreement with PFb = F plunger + W r − W f − Phd (3)
the measured static PRLs at the upstroke and downstroke,
respectively. In combination with fluid density, pump depth and On the contrary, at the end of upstroke (point d), where the
its size, these two measured static PRLs are applied to calculate suspension point achieves the largest displacement, its velocity
the pump intake pressure. Finally, the pump submergence is approaches zero. Since the standing valve is open and the
determined and converted to obtain the dynamic fluid level. In traveling valve remains closed, the weight of the fluid on the
addition, a field case is presented to demonstrate the design and full plunger area becomes the main load. Because the standing
application of the newly developed technique. valve is open, the sucker-rod and plunger undergo the intake
pressure other than the buoyancy force of the fluids. So, the
PRL at point d, PFd , is
Theoretical Model PFd = Wl + Fplunger + Wr + Phu − Pi (4)

[ ]
Pi = ρ o g ( L − hc ) + p casing A p (5a)
Determination of average PRLs
W l = (A p − Ar )L ρ l g (5b)
Phu = (Ap − Ar ) p oil
It is assumed that the sucker-rod string is a uniform diameter
(5c)
rod and not twisted in the tubing. Also, it is reasonably assumed
that the liquid above the plunger can be considered as a single where Wl is the weight of the fluid on the cross-sectional area
phase fluid. Figure 1 is a typical surface dynamometer card with
gas effect. It has been found from the measured dynamometer of the plunger, N; Pi is the force imposed by the intake
card that there exists two turning points at the end of the pressure on the bottom of the plunger, N; Phu is the force
upstroke (points a and b) and downstroke (points d and e),
respectively. imposed by the THP at the upstroke, N; pcasing is the casing
As shown in Figure 2, at point a, an upstroke begins and the
head pressure (CHP), Pa; and hc is the dynamic fluid level, m,
horse head is at its lowest point and the velocity of the
suspension point is close or equal to zero. At this moment, the assuming the fluid in the annulus is crude oil.

2
 Similarly, the PRL at point e, PFe , can be obtained as less than 5% account for 65.64%. It is also shown in Figure 4(b)
that the mean relative error is 1.54% and the standard deviation
PFe = Wl − Fplunger + Wr + Phu − Pi (6)
is 9.78%.
Combining Equations (1) and (3) yields In comparison of Figure 3(a) with Figure 4(a), it is found that
PFa + PFb (7) the difference between the measured static PRLs and the
Fdown = = Wr − W f − Phd
2
calculated average PRLs, i.e., Fup and Fdown , is relative large
where Fdown is the average PRL between points a and b as
at the downstroke. This is mainly due to the fact that, at the
shown in Figure 1. downstroke, existence of gas prevents the traveling valve from
Similarly, combining Equations (4) and (6) yields completely opening, and thus the measured static PRL is
PFd + PFe
Fup = = Wr + Wl + Phu − Pi (8) smaller. On the other hand, the leakage of pump makes the
2 measured static PRLs larger at the downstroke. In addition, the
where Fup is the average PRL between points d and e, N, as inappropriate stop position and timing is more sensitive to the
shown in Figure 1. In practice, if the load due to acceleration is measured static PRLs at the downstroke than at the upstroke. In
practice, the static PRLs can be readily obtained from the
considered, Fdown and Fup can also be determined in a similar
dynamometer card. Therefore, the static PRLs at the upstroke
way; however, the weighted-factor will be changed. As shown and downstroke obtained from the measured dynamometer
in Figure 1, the two horizontal lines represent the average PRL, cards can be used as the values of Fup and Fdown .
Fdown , between point a and b and the average PRL, Fup ,
between point d and e. Determination of dynamic fluid levels
It is also found that the two calculated average PRLs are
very close to the measured static PRLs, respectively, as shown Since the dynamic fluid level is dependant on the reservoir
in Figure 1. Therefore, the measured static PRLs at the upstroke performance, the well deliverability can be related to the work
efficiency of the polished-rod, which is equal to the enveloped
and downstroke can be used to determine Fup and Fdown ,
area in a dynamometer card. Accordingly, the dynamic fluid
respectively. In practice, if points a and b are overlapped, the level imposes a significant effect on the shape of the
calculated Fdown is the same as the static PRLs at the dynamometer card.
The pump submergence is defined as the plunger depth
downstroke. Similarly, the calculated Fup is the same as the
minus the dynamic fluid level. The pressure due to the pump
static PRLs at the upstroke if the points d and e are overlapped. submergence termed as the intake pressure ( Pi ), which imposes
a direct effect on the static PRL at the upstroke. As long as the
Comparison of the calculated average static PRL is known at the end of the upstroke, the pump intake
PRLs and the measured static PRLs pressure can be determined[17]. According to Equation (5a),
In 2006, 203 measured dynamometer cards for 11 wells are L − hc can be obtained and easily converted to the dynamic
randomly chosen from more than 800 wells in the Qinghai fluid level.
Oilfield, CNPC, to test the technique developed in this study.
The static PRLs are measured when the pump stops at the Combining Equations (2a-2c) and W f = ρ l Wr , Equation (7)
upstroke or downstroke till pumping unit becomes stable. ρr
In general, there exists a small deviation for the static PRLs can be rewritten as
measured at the upstroke and downstroke, respectively. This is
mainly ascribed to the inappropriate stop for the upstroke and Fdown + Ar Poil (9)
Wr =
downstroke or to the malfunction of the dynamometer and well. ρ
1− l
After analyzing all the dynamometer cards, 191 measured static ρr
PRLs for the upstroke and 163 measured static PRLs at Similarly, combining Equations (5a-5c) with consideration
of Wl = ρ l gLAp − ρ l Wr , Equation (8) can be rewritten as
downstroke are found to be credible.
The relative errors between the measured static PRLs and the ρr
calculated average PRLs versus the well sample numbers at the
ρ
upstroke and the normal distribution for the relative error are Fup = Wr + ρ l gLA p − l Wr + (Ap − Ar ) poil
plotted in Figures 3(a) and (b), respectively. Here, the relative ρr (10)
error is defines as the difference between the measured static − [ρ o g ( L − hc ) + pcasing ]Ap
PRL and the calculated average PRL divided by the measured Hence, the dynamic fluid level, h , can be obtained
static PRL. It is found from Figure 3(a) that, for the credible c

191 measurements for the static upstroke PRLs, the samples Fup − Fdown poil − pcasing ⎛ ρ ⎞ (11)
hc = − + ⎜⎜1 − l ⎟⎟ L
with relative error of less than 7% account for 92.14%, while ρ o gAp ρo g ⎝ ρo ⎠
the samples whose relative error are less than 5% account for For field applications, when the difference between the THP
79.58%. In addition, it can be seen from Figure 3(b) that the and CHP is very small and the fluid in the annulus is the same
mean relative error is 2.14% and the standard deviation is as that inside the tubing, Equation (11) can be simplified as
6.68%. Fup − Fdown
The relative errors between the measured static PRLs and the hreduced = ≈ hc (12)
ρ o gAp
calculated ones versus the well sample numbers at the
downstroke and the normal distribution for the relative error are where hreduced is the reduced dynamic fluid level, m.
depicted in Figures 4(a) and (b), respectively. It is found from It can be seen from Equation (12) that only four variables are
Figure 4(a) that, for the credible 163 measurements of the static needed to determine the dynamic liquid level, among which
PRLs at the downstroke, samples with relative error of less than Fdown and Fup can be obtained from a measured dynamometer
7% account for 79.75%, while samples whose relative error are

3
card directly. This method provides a pragmatic approach for NOMENCLATURE
determining the dynamic fluid level by analyzing the measured
dynamometer card, especially when the automatic monitoring Notations
technique has been used for measuring the dynamometer cards. Ap = cross-sectional area of the plunger, m2;
If other parameters, such as poil and p casing , can be obtained by Ar = cross-sectional area of the sucker-rod, m2;
using the automatic monitoring system, the real-time dynamic fw = water-oil ratio;
fluid level can be determined more accurately.
Fdown = average PRL between points a and b, N;
Fplunger = friction between plunger and barrel, N;
Case Study FSdown = measured static PRL at the downstroke, N;
The technique developed in this study has been successfully F Sup = measured static PRL at the upstroke, N;
applied to determine the dynamic fluid levels for more than 400
wells in the Qinghai Oilfield. For demonstration purpose, the Fup = average PRL between points d and e, N;
measured dynamic fluid levels for 16 wells are listed Table 1. g = gravitational acceleration, m/s2;
Also, this table lists all the physic properties for the pump,
fluids and wells. The density of the formation water is 1005 hc = dynamic fluid level, m;
kg/m3. hreduced = reduced dynamic fluid level, m;
It can be seen from this table that the relative error between
the measured dynamic fluid level and the calculated value is htest = measured dynamic fluid level, m;
less than 6.07%. It is also can be seen that the dynamic fluid L = length of the sucker rod string, m;
levels determined by using Equations (11) and (12) are very PFa = PRL at point a, N;
close to each other if the difference between the THP and the
CHP is small. This means that the dynamic fluid level can be PFb = PRL at point b, N;
determined with the simplified method for engineering PFd = PRL at point d, N;
application purpose as long as the difference between the THP
and the CHP is small enough and the fluid in the annulus is the PFe = PRL at point e, N;
same as that inside the tubing. p casing = casing head pressure (CHP), Pa;
It should be noted that this technique can also be used to
measure the dynamic fluid level for the wells where the acoustic Phd = forces imposed by the tubing head pressure
method or other similar methods are not applicable. For at the downstroke, N;
example, this method is still applicable when the casing
produces crude oil because of its containing a large amount of Pi = intake pressure on the bottom of plunger, N;
gas. Phu = forces imposed by the tubing head pressure
at the upstroke, N;
poil = tubing head pressure, Pa;
Conclusions
Wf = buoyancy force on the rods, N;
A pragmatic technique is successfully developed and applied
Wl = weight of fluid worked on the cross-
to determine the dynamic fluid level by analyzing the measured
dynamometer cards. Theoretically, a mathematical model is sectional area of the plunger, N;
developed to calculate the average PRLs for the two turning Wr = weight of the rods in air, N;
points at the end of the upstroke and downstroke, respectively.
In practice, the average PRLs can be directly obtained from the
dynamometer cards and then used to determine the dynamic Greek symbols
fluid level. In addition to accurate determination of the real-time ρl = fluid density, kg/m3;
dynamic fluid level, this method greatly reduces the production
maintenance cost. The method developed in this paper has been ρo = crude oil density, kg/m3;
applied to more than 400 wells. It is shown from the field ρr = density of sucker rod, kg/m3;
applications that the average error for the dynamic fluid level is
within 50 m. In particular, this technique provides a real-time ρw = water density, kg/m3.
approach for determining the dynamic fluid level in those wells
where the automatic monitors are installed and operated in real-
time. A PC-based software for determining the dynamic liquid REFERENCES
levels is also developed.
1. Lieberman, S., Automated Continuous Fluid Level
Monitoring; paper SPE 93832, presented at the 2005 SPE
Production and Operations Symposium, Oklahoma, OK,
Acknowledgments April 17-19, 2005.
The authors would like to acknowledge Hailang Zhang, 2. Sun, S., Song, J., Zhang, Z., Bian, L., Li, Y., and Yan, J.,
Heng Li, and Shengnan Chen for their valuable technical Reservoir Pressure Obtained from Dynamic Fluid Level;
discussions. Petroleum Geology & Recovery Efficiency, Vol. 11, No.4,
pp. 49-50, August 2004.
3. Pan, Z., Zhu, G., and Fu, G., Application in Dynamic Oil
Level Measurement; Microprocessors, pp. 62-64,
February 2000.

4
4. Sun, X., Huang, G., Li, A., Zhang, C., Guo, Y., and Li, X., Failure in High Water-cut Stage; China Petroleum
Derivation and Application of the Equation about Dynamic Machinery, Vol. 31, No. 7, pp.39-42, July 2003.
Fluid Levels of Oilwell; Henan Petroleum, Vol. 18, No.3, 12. Schirmer, P., Gay, J.C., and Toutain, P., Use of Advanced
pp. 54-55, June 2004. Pattern-Recognition and Knowledge-Based System in
5. Nind, T.E.W., Principles of Oil Well Production; McGraw- Analyzing Dynamometer Cards; paper SPE 22305,
Hill, New York, 1981. presented at the Sixth SPE Petroleum Computer
6. von Pattay, P.W., Strasser, R., and Seitzjef, A., Innovative Conference, Dallas, TX, June 17-20, 1991.
Approach for Automatic Real-Time Monitoring of the 13. Chai, M., Development of a Real-time Monitoring
Fluid Level in Wells Without Returns to Surface During Pumping Well Electric Power System; Journal of
Workover Operations-A Successful Field Test; paper SPE Jianghan Petroleum Institute, Vol. 25, No. 3, pp. 120-121,
100278, presented at the SPE Europe/EAGE Annual September 2003.
Conference and Exhibition, Vienna, Austria, June 12-15, 14. Main, M.A., Petroleum Engineering Handbook for the
2006. Practical Engineer, Vol. II; PennWell Publishing
7. Fu, Y., Yu, X., and Lin, Z., Performance Analysis of SL- Company, Tulsa, OK, 1991.
MSL-I Automatic Level Monitor; Journal of Daqing 15. Zhang, Q., Production Engineering and Design; Press of
Norma1 University, Vol. 26, No. 2, pp. 35-38 , April 2006. China University of Petroleum, Dongying, China,
8. Nolen, K.B. and Gibbs, S.G., Measurement and September 2000.
Interpretation of Fluid Levels Obtained by Venting Casing 16. Schnitman, L., Albuquerque, G.S., Corrêa, J.F., Lepikson,
Gas; paper SPE 38791, presented at the 1997 SPE Annual H., and Bitencourt, A.C.P., Modeling and Implementation
Technical Conference and Exhibition, San Antonio, TX, of a System for Sucker Rod Downhole Dynamometer Card
October 5-8, 1997. Pattern Recognition; paper SPE 84140, presented at the
9. Wang, H., Wang, Y., and Han, L. Oil Production Obtained SPE Annual Technical Conference and Exhibition, Denver,
by Using Dynamometer Cards; Oil-Gas Field Surface CO, October 5-8, 2003.
Engineering, Vol. 23, No. 3, pp. 29-29 , March 2004. 17. Wang, Y., Wang, H., Zhang, C., and Lai, M., Computing
10. Everitt, T.A. and Jennings, J.W., An Improved Finite- Static Hanging Load of Sucker Rod in Pumping Well by
Difference Calculation of Downhole Dynamometer Cards Means of Leakage Curves; Journal of Xi’an Petroleum
for Sucker-Rod Pumps; SPE Production Engineering, Vol. University (Natural Science Edition), Vol. 20, No. 2, pp.
7, No. 1, pp. 121-127, February 1992. 58-60 , March 2005.
11. Gao, J. and Yu, T., The Research and Analysis of
Dynamometers: Characteristics and Downhole Tools-

Table 1 Physical properties of the oilwells and comparison between the measured dynamic fluid levels and the calculated data
Plunger htest − hc hc − hreduced
No.
Well Pump fw THP CHP ρo depth
Fup Fdown htest hc hreduced
htest hc
Name diameter (MPa) (MPa) (N) (N) (m) (m) (m)
(kg/m3) (m) (%)
(mm) (%)
1 10-10 44 0.99 1.3 1.5 787 1504.2 55.62 34.32 1486 1435.0 1425.2 3.43 0.68
2 313 57 0.52 1.0 1.0 791 1505.4 56.43 30.99 1105 1104.3 1127.5 0.06 -2.10
3 330 57 0.05 1.3 1.3 793 1452.6 55.30 45.36 517 548.4 495.0 -6.07 9.74
4 332 57 0.66 2.0 1.6 791 1502.1 69.87 38.38 1317 1287.9 1350.6 2.21 -4.87
5 3321 44 0.50 1.0 1.0 785 1649.5 61.03 38.77 1614 1607.2 1669.1 0.42 -3.85
6 338 44 0.65 1.4 0.8 785 1652.2 63.57 41.57 1456 1512.9 1590.6 -3.91 -5.14
7 339 70 0.28 1.2 1.2 803 1199.8 51.92 40.00 * 357.8 367.4 -2.68
8 3393 57 0.10 1.4 0.8 787 1603.0 65.75 47.30 898 873.2 911.8 2.76 -4.42
9 3394 57 0.17 0.7 0.0 795 1410.0 59.97 38.27 926 969.6 1045.2 -4.71 -7.80
10 341 57 0.65 0.8 0.8 788 1584.1 63.63 41.06 949 914.7 971.6 3.61 -6.22
11 350 44 0.26 0.8 0.6 789 1556.3 59.76 41.04 1491 1462.7 1486.3 1.90 -1.61
12 351 70 0.87 1.8 0.0 797 1357.2 44.56 34.26 337 0 279.2
13 3526 44 0.09 1.6 1.3 785 1653.1 61.78 43.45 1516 1498.1 1527.5 1.18 -1.96
14 3527 70 0.12 1.6 0.0 795 1405.5 62.25 38.31 * 604.3 774.1 -28.10
15 359 44 0.31 1.7 1.5 795 1403.8 60.08 40.98 1401 1467.1 1490.1 -4.72 -1.57
16 364 57 0.35 1.4 0.0 785 1655.6 67.74 41.77 1046 1030.0 1204.5 1.53 -16.94

Note: * The casing produces crude oil due mainly to its containing a large amount of gas.

5
 70

d
60
FSup

Fup
Load (kN)

c
50

e
b
Fdown

FSdown
40
f
a

30
0.0 0.5 1.0 1.5 2.0 2.5 3.0

Stroke (m)

Figure 1 Typical dynamometer card together with the calculated Fup and Fdown as well as the measured static PRLs

(a) (b)
Figure 2 Schematic of the downhole pump at (a) point a and (b) point b.

6
 60
Upstroke

40
Relative Error (%)

20

-20

-40

-60
0 20 40 60 80 100 120 140 160 180 200 220

Sample Well Number

(a)

M e a n = 2 .1 4 % , S td D e v = 6 .6 8 %
14

12

10

0
-3 5 -3 0 -2 5 -2 0 -1 5 -1 0 -5 0 5 10 15 20 25 30 35
R e la tiv e E rr o r ( % )
(b)
Figure 3 (a) Relative errors between the measured static upstroke PRL and the calculated Fup versus sample numbers; (b) Normal
distribution of the relative errors between the measured static upstroke PRL and the calculated Fup .

7
 60
Downstroke

40
Relative Error (%)

20

-20

-40

-60
0 20 40 60 80 100 120 140 160 180 200 220

Sample Well Number

(a)

M e a n = 1 .5 4 % , S td D e v = 9 .7 6 %
14

12

10

0
-3 5 -3 0 -2 5 -2 0 -1 5 -1 0 -5 0 5 10 15 20 25 30 35

R e la tiv e E rro r ( % )
(b)
Figure 4 (a) Relative errors between the measured static downstroke PRL and the calculated Fdown versus sample numbers; (b)
Normal distribution of the relative errors between the measured static downstroke PRL and the calculated Fdown .