Helping Engineers Control Corrosion & Cracks

BEASY SOFTWARE & SERVICES

Assessing Interference Between Sacrificial Anodes on Anode Sleds

Tim Froome
CM
Assessing Interference Between BEASY Ltd Anodes on Anode Sleds
Sacrificial
Ashurst Lodge
Southampton, SO40 7AA
Tim Froome
England
CM BEASY Ltd
Ashurst Lodge
Southampton, SO40 7AA
John MW Baynham
England
CM BEASY Ltd
[email protected]
ABSTRACT
John MW Baynham
Design of sacrificial anode cathodic protection CM BEASY
systemsLtd sometimes involves use of an array of anodes
mounted on a sled, which is connected to the structure by cable to provide a return path. Two factors
can limit the effectiveness of sled-mounted anodeABSTRACT
arrays: firstly the IR drop along the return path cable
can significantly reduce the driving potential; secondly close proximity of anodes can reduce the output
Design of sacrificial
of the array anodewhat
to well below cathodic
wouldprotection systems
be expected if the sometimes
anodes wereinvolves use of an array of anodes
well separated.
mounted on a sled, which is connected to the structure by cable to provide a return path. Two factors
can
Thelimit theofeffectiveness
effect of sled-mounted
IR drop can easily anode
be estimated usingarrays:
knownfirstly
cablethe IR drop along
resistance the return current.
and anticipated path cable
can significantly reduce the driving potential; secondly close proximity of anodes can reduce the output
ofHowever
the arraywith
to well below
respect to what
anode would be available
output, expected data
if thefor
anodes were to
resistance well separated.
ground is limited to fairly simple
groupings of anodes.
The effect of IR drop can easily be estimated using known cable resistance and anticipated current.
For a complex anode sled design it is only practically possible to determine the theoretical maximum
However
output bywith
userespect to anode modelling
of mathematical output, available datausing
performed for resistance to ground
computer-based is limited
simulation to fairly simple
techniques.
groupings of anodes.
First, accuracy of the simulation method is established by comparison with exact analytical methods
For
andaapproximate
complex anode sled design
formulas used byitindustry.
is only practically possible to determine the theoretical maximum
output by use of mathematical modelling performed using computer-based simulation techniques.
Secondly, accuracy is assessed of approximate formulas when applied to anode geometries different
First,
from accuracy of the the
those for which simulation
formulasmethod is established by comparison with exact analytical methods
were derived.
and approximate formulas used by industry.
Finally, resistance to ground is determined for three different sled designs.
Secondly, accuracy is assessed of approximate formulas when applied to anode geometries different
from
Key those
words:foranode,
which resistance
the formulas were derived.
to ground, analytic, approximate, simulation, sled, interference ratio
Finally, resistance to ground is determined for three different sled designs.

Key words: anode, resistance to ground, analytic, approximate, simulation, sled, interference ratio

Ashurst Lodge, Ashurst, Southampton, SO40 7AA, UK Tel: +44 (0)238 029 3223 Fax: +44 (0)238 029 2853
25 Bridge St, Billerica, MA 01821, USA Tel: +1 978 667 5841 Fax: +1 978 667 7582

Email: [email protected] Website: www.beasy.com

 INTRODUCTION

Available methods for prediction of anode resistance include analytical formulae (generally exact for the
assumed conditions), approximate formulae (generally reasonably accurate for the assumed
conditions), and methods based on use of numerical simulation (which can be very accurate, and which
can use virtually any anode geometry and any number of anodes).

Analytical formulae are known for a range of specific, relatively simple, geometries, such as long thin
wires and spheres.

Approximate formulae have been derived by a number of authors for a bigger range of geometries,
including multiple anodes. One method of calculation used is the average potential method, which
according to Dwight1 is correct to within 2 or 3 parts in 1000 for a long straight wire, and within less than
1 part in 100 for a cylinder of the proportions of a ground rod (he gives an example of a ground rod 3 m
(10 ft) long and 20 mm (0.75 in) diameter, i.e. length/radius = 320).

A range of simulation methodologies could be applied to this problem, of which the boundary element
method is applied in this work, using the commercial software package BEASY(†). The simulation uses
an accurate model of the anode array to determine resistance to ground, taking into account as
required effects of the sea surface or seabed.

Although resistance to ground provides a measure of the ability of the sled to deliver current, a more
realistic approach is to take into account proximity of the sled to the structure to be protected (and
possibly to other sleds) and to include resistance of the return path. Although not normally necessary
(since its effect is minor) a model of the structure of the sled itself may be created. Simulation is able to
take into account polarization of the surfaces, coating breakdown, accumulation of calcareous deposits,
and change of anode shape (and hence resistance to ground) as the anode is consumed.

However the main objective of this paper is to quantify interference effects between sled-mounted
sacrificial anodes and thereby provide data which can be used at the early design stage. Therefore this
paper concentrates on calculation of resistance to ground of anodes on the sled, and does not take into
account the effects of polarization, effects of change of anode shape during lifetime of the anode, etc.

Assessment of Accuracy of The Available Methods for Calculation of Anode Resistance

A number of anode types are considered, and where possible each of the three available methods is
applied to determine anode resistance to ground. The form of any analytic or approximate formula
which has been used is shown, with identification of the assumed conditions for which the formula was
derived.

Because the accuracy of the analytic and approximate methods is only guaranteed for the assumed
geometric conditions, parameter studies are generally performed to help assessment of loss of
accuracy when the formulas are applied to other geometry.

The simulation result is obtained by representing the surfaces of the anode and surfaces of a far-away
“bounding box” using boundary elements, and applying potential values to each of these surfaces. The
resulting current flowing from the anode to the “infinite” boundary, together with the applied potential
difference, directly identifies the resistance to ground.

For simplicity the resistivity of the electrolyte (ground or seawater) is taken to be 1 Ohm-m.

†
Trade name
 RESULTS

This section shows results of the investigations for nine different examples of generally increasing
complexity.

Example 1: Single Vertical Ground Rod of Length “L”, Extending to Depth “L” in the Ground

An exact formula can be obtained for a rod for which “radius is much smaller than the length”, using an
exact equation which Dwight1 attributes to Hallen2. The resulting formula, using terms included in
reference 1, is:
ρ  1 1.22741 2.17353 11.0360 
R=  + + +  (1)
2ΠL λ 4λ2 8λ3 16λ4 
Where:
• L is the length of the vertical rod
• λ = loge(2L/a),
• a is the radius of the rod
• “ρ” is the ground resistivity

Subsequently we use “ln” as shorthand for “loge”.

In equation 20a in reference 2, Hallen shows the exact expressions which lead to the terms involving λ
are as follows:
 2 
 (8 ln(2) 2 − 16 ln(2) + 16 − Π 2 ) 
 1 + 4(1 − ln(2)) + 3 +
B  (2)
λ 4λ 2
8λ 3
16λ 4 
 
 
Where:
1
B = −16 ln(2) 3 + 48 ln(2) 2 − 96 ln(2) + 96 − 4Π 2 (1 − ln(2)) − 20∑1
∞
(3)
n3

Expressions (3) and (4) are used in formula (1) in this work to determine the analytic result.

An approximate formula applicable to this case is equation (13) in reference 1, which Dwight identifies
as “the resistance to ground of a vertical ground rod of depth L”. (Note it is clear in reference 1 that the
ground rod extends from the ground surface.) The formula is as follows:
ρ   4L  
R=  ln  − 1 (4)
2ΠL   a  
In the above equation terms have the same meaning as before.

Formula (4) is recommended by some authorities (eg DNV-RP-B4013) for application to long slender
stand-off anodes for which length is greater than 4 times the radius.

Comparison of results.
We have determined resistance to ground using each of the above three techniques, for a range of
values of radius, with rod length 1 m, and ground resistivity 1 Ohm-m.

Results of these calculations are shown in Table 1.

 Table 1
Resistance to Ground Calculated using the Analytic Formula for a Long Thin Ground Rod, the
Approximate Formula for a Long Thin Ground Rod, and Numerical Methods, for Various
Length/Radius Ratios
%error of
%error of %error of %error of
length / "approx"
R "analytic" R "approx" R "sim" "sim" cf "approx" "analytic"
radius cf
"analytic" cf "sim" cf "sim"
"analytic"
10000000 2.624778 2.6267553 2.624672 0.0753 -0.0041 0.0794 0.0041
1000000 2.257946 2.2602875 2.257795 0.1037 -0.0067 0.1104 0.0067
100000 1.890954 1.8938197 1.890645 0.1515 -0.0163 0.1679 0.0163
10000 1.523676 1.5273519 1.522939 0.2412 -0.0484 0.2897 0.0484
1000 1.15581 1.1608841 1.153875 0.4390 -0.1674 0.6074 0.1677
100 0.786448 0.7944163 0.782656 1.0133 -0.4821 1.5026 0.4844
10 0.411604 0.4279485 0.417545 3.97 1.44 2.49 -1.42
5 0.295298 0.3176307 0.317425 7.56 7.49 0.06 -6.97
2 0.136059 0.1717985 0.202382 26.3 48.7 -15.1 -32.8
1 0.027043 0.0614807 0.133582 127.3 394.0 -54.0 -79.8

Discussion of example 1.
Under conditions closest to those for which it was derived (i.e., Length/radius>> 1, eg L/r=107 in Table
1) it can be assumed that the analytic formula is highly accurate, and accuracy of the approximate and
simulation results should be measured against it.

Thus the column headed “%error of ‘approx’ cf ‘analytic’” shows the percentage error of the
approximate method when compared with the analytic result. Similarly the column headed “%error of
‘sim’ cf ‘analytic’” shows the percentage error of the simulation method when compared with the
analytic result.

It can be seen for the case when Length/radius = 107 that whereas the approximate method is
reasonably accurate (to within 0.08%) the simulation result is extremely accurate (to within 0.0041%).

Viewing other results in the table, three points should be considered:

1. As length/radius reduces, the assumptions made when deriving the analytic become
less representative of the geometry considered, and therefore the analytic solution is no
longer exact.

2. Dwight states that the approximate “method is correct to within 2 or 3 parts in 1000 for a
long straight antenna wire, and within less than 1% for a cylinder of the proportions of a
ground rod.” He gives as an example a ground rod with length/radius = 320.

It can be seen from the table that at Length/radius=10000 the approximate method has
error of 2.4 parts in 1000 when measured against the “analytic” (Column headed “%error
of ‘approx’ cf ‘analytic’”).

Interpolating (linearly) to ratio Length/radius = 320, the approximate method has error
0.87 parts in 100 when measured against the “analytic”
 These two observations corroborate the error estimates made by Dwight, and show that
he was comparing against the analytic result (which we know is not exact for smaller
values of length/radius, but have no measure of the loss of accuracy).

3. Because the simulation method is general in the way it applies to geometry, there is no
reason to expect accuracy any different from that shown for the case when
Length/radius =107.

Therefore it is appropriate to measure accuracy against the simulation result. This is

done for the approximate method in the column headed “%error of ‘approx’ cf ‘sim’”, and
for the analytic method in the column headed “%error of ‘analytic’ cf ‘sim’”.

It can be seen from the table that when measured against the simulation result, the
approximate method has error of 2.9 parts in 1000 at length/radius = 10000, and (by
interpolation) has an error of 1.28 parts in 100 at length/radius = 320.

Example 2: A Sphere Well Away from the Ground Surface

To test the simulation result for different types of geometry, we now consider an example which is not
long and thin, and for which an analytic solution is available.

An exact formula is known for an isolated sphere as follows:

ρ
R= (5)
4Π a
Where:
• a is the radius of the sphere
• “ρ” is the resistivity of the electrolyte

Comparison of results.
We have determined resistance to ground using analytic and simulation methods, with a sphere of
radius 1 m, and ground resistivity 1 Ohm-m. Results of these calculations are shown in Table 2.

Table 2
Resistance to Ground of a Sphere, Calculated using the Analytic Formula and Simulation
%error of "sim"
R "analytic" R "sim" cf "analytic"
0.07957747 0.07957222 -0.006599606

Discussion of example 2.
It can be concluded from Table 2 that the simulation method is accurate for geometry which is not long
and thin.
Example 3: A Pair of Ground Rods

To test the representation of interference effects we first consider a pair of electrically connected
ground rods. An approximate formula for this case (a pair of cylinders with length/radius >>1) was
published by Dwight as equation 19 in reference 1 as follows:
ρ  4L ( 2 L + s 2 + 4 L2 s s 2 + 4 L2 
R= ln( ) − 1 + ln( )+ −  (6)
4Π L  a s 2L 2L 
Where:
• a is the radius of the rods
• L is the length of the rods, which extend vertically from the ground surface
• s is the distance between the centre lines of the rods
• “ρ” is the ground resistivity

Long very thin ground rods.

We have used equation (6) to determine resistance to ground of a pair of rods of length 7 m and radius
0.0001 mm in ground of resistivity 1 Ohm-m, for a range of separation distances, and have measured
the error of the result by comparing with resistance obtained using simulation. The change of error with
separation distance (see Figure 2) shows that the error is less than about 0.1% if the separation
distance is greater than about 2 m.

Long thin ground rods.

Similarly we have used equation (6) to determine resistance to ground of a pair of rods of length 7 m
and radius 10 mm in ground of resistivity 1 Ohm-m, for a range of separation distances, and have
measured the error of the result by comparing with resistance obtained using simulation. The change of
error with separation distance (see Figure 2) shows that the error is less than 1% if the separation
distance is greater than about 2 m.

Short fat ground rods.

Finally we have used equation (6) to determine resistance to ground of a pair of rods of length 1 m and
radius 100 mm in ground of resistivity 1 Ohm-m, for a range of separation distances, and have
measured the error of the result by comparing with resistance obtained using simulation. The change of
error with separation distance (see Figure 3) shows that the error is less than about 2.4% if the
separation distance is greater than about 2 m.

Discussion of example 3.
These results show that the error in the approximate equation increases both as length/radius
decreases, and as separation of the two anodes decreases.

For ground rods of similar proportions to those of sacrificial anodes (as in Figure 3) the approximate
method is roughly 2.5% in error even at large spacing, but the error increases very sharply at spacing
less than about 2 m, with the approximate method over-predicting the resistance to ground of the pair
of rods.
 Two ground rods of length 7m and diameter 0.0002 mm

0.14
Percentage error of resistance calculated

0.12
using the approximate method

0.1

0.08

0.06

0.04

0.02

0
0 2 4 6 8 10 12 14 16 18 20
Distance between centre lines of ground rods (metres)

Figure 1: Variation of Error (of The Resistance to Ground Calculated Using The Approximate
Equation) with Separation of Two Ground Rods of Length 7 m and Diameter 0.0002 mm

Two ground rods of length 7m and diameter 20 mm

1.2
Percentage error of resistance calculated

1
using the approximate method

0.8

0.6

0.4

0.2

0
0 2 4 6 8 10 12 14 16 18 20
Distance between ground rods (metres)

Figure 2: Variation of Error (of The Resistance to Ground Calculated Using The Approximate
Equation) with Separation of Two Ground Rods of Length 7 m and Diameter 20 mm
 Two ground rods of length 1m and diameter 200 mm

6
Percentage error of resistance calculated

5
using the approximate method

0
0 5 10 15 20 25 30
Distance between centre lines of ground rods (metres)

Figure 3: Variation of Error (of The Resistance to Ground Calculated using the Approximate
Equation) with Separation of Two Ground Rods of Length 1 m and Diameter 200 mm

Example 4: A Square Grid of Ground Rods

As the next step in testing the representation of interference effects we consider a regular square grid
of ground rods as studied by Dwight, who published results of the approximate method in graphical
form in figure 5 in reference 1. Data extracted from his graph is reproduced in this paper in Figure 4,
which also shows results of simulation for a number of cases. The graph shows variation with number
of rods in the array of the following ratio (which is a convenient measure of interference, and is called
the “interference ratio” in the remainder of this paper):

    

interference ratio          (7)
!      

Each ground rod of diameter 20 mm (0.75 in) extends 3 m (10 ft) vertically from the ground surface,
and the rods are arranged on a regular square grid at various spacing between rods, roughly 1.5, 3,
6.1, 9.1, 12.2, 18.3, 24.4, and 30.5 m (exactly 5, 10, 20, 30, 40, 60, 80, and 100 ft).

Simulation results (shown in Figure 4 as square symbols) were obtained for 4, 16, 36 and 64 rods, at
spacing of roughly 1.5, 3, 6.1, and 30.5 m (exactly 5, 10, 20, and 100 ft).

Discussion of example 4.
It was noted by Dwight that for a large number of ground rods the results shown by solid lines are
approximate. This is particularly evident at smaller spacing, for example with roughly 1.5 m (exactly 5 ft)
spacing with 16 rods the solid curve is in error by 20% when compared with the simulation result. The
accuracy of the solid curves is somewhat variable, but appears to be reasonably accurate at spacing
6.1 m (20 ft) or greater.
Figure 4: This Figure Reproduces Data Published by Dwight as "Figure 5" in Reference 1 (Solid
Curves) and Data Obtained using Simulation (Squares). Colors indicate spacing as follows: Red
= 1.5 m (5 ft), Green = 3 m (10 ft), Blue = 6.1 m (20 ft), Cyan = 9.1 m (30 ft), Magenta = 12.2 m (40
ft), Yellow = 18.3 m (60 ft), Orange = 24.4 (80 ft), Purple = 30.5 m (100 ft). The “Ratio” is the
Interference Ratio Defined in Equation 7 Above.
Example 5: A Single “Cuboid” Anode Well Away From the Seawater Surface

We now start to consider typical sacrificial anode shapes used in marine applications. In this example
we establish the resistance to ground of a single isolated anode well away from any structure, from the
seabed and from the seawater surface.

Dwight published equation 12 in reference 1 and identifies it as “the resistance of a buried straight wire
of length 2L, no part of which is near the surface of the ground”. The formula is as follows:
ρ   4L  
R=  ln  − 1 (8)
4ΠL   a  
Where:
• L is half of the length of the anode
• a is the radius of the rod
• “ρ” is the ground resistivity

Dimensions (in m) of the anode with trapezoidal cross-section are shown in Figure 5. The ends of the
anode are parallel. We have used seawater resistivity 1 Ohm-m.

Figure 5: Dimensions of the Sacrificial Anode used in Example 5, 6, 7 and 9

Formula (8) can be applied to an anode of this shape (albeit with increased approximation) by
determining the “equivalent” radius using:
c
a= (9)
2Π
Where:
• a is the equivalent radius of the anode
• c is the circumference of the anode

For the anode shown in Figure 5 the circumference is 0.907056 m, and the equivalent radius is 0.14436
m.

Results for example 5.

Simulation produced contours of current density as shown in Figure 6, and resistance of the single
anode was determined to be 0.1641 Ohms. The resistance calculated using formula (8) is 0.162011
Ohms.
Discussion of example 5.
The error of the approximate calculation performed using “Dwight equation 12” is about 1%. This
confirms that the most appropriate formula for an isolated anode (i.e. well away from structure, seabed
or sea surface) is formula (8) above. Note carefully the definition of “L” in formula (8)!

This is in apparent contradiction to the advice in document DNV-RP-B4013, which is that formula (4)
above should be used. If formula (4) is used, the resistance is calculated to be 0.206138 Ohms, which
is in error by nearly 26%.

The next example investigates the appropriateness of formula (4) above (i.e. Dwight equation 13) by
solving for resistance of a single anode at various distances from an insulating surface (such as the sea
surface).

Figure 6: Contours of Current Density on the Surface of the Single Anode. Dark Blue is Smallest
Current Density, Red is Biggest Current Density.

Example 6: A Single Cuboid Anode at Various Distances from the Seawater Surface

To establish the appropriateness of Dwight equation 12 or Dwight equation 13 for use with sacrificial
anodes in seawater, we now position the anode (dimensioned as in Figure 5) at various separation
distances from an insulating surface (which could be regarded as the sea surface or a perfectly coated
surface) as shown in Figure 7.

Figure 7: Showing End View of the Anode at Some Distance from an Insulating Surface
Results for example 6.
Simulation results showing variation of resistance (with separation distance) are shown in Figure 8 and
in more detail in Figure 9.

In these figures:
• the red line shows the resistance to ground obtained by simulation
• the green line shows resistance calculated using Dwight equation 12
• the dark blue line shows resistance calculated using Dwight equation 13
• the pale blue line shows resistance calculated using the DNV formula4 for a “long flush mounted
anode”
The effect of separation distance on current density on the anode surfaces is shown in Figure 10.

Discussion of example 6.
The simulation shows:
• At zero separation distance the back face of the anode does not deliver current, and simulation
shows the resistance to be 0.28 Ohms, which is significantly less than the 0.36 Ohms obtained
using the DNV formula for a long flush mounted anode4. The DNV formula over-predicts the
resistance by 27%.
• For distance from zero to about 0.67 m the resistance is bigger than that calculated using
Dwight equation 13. For example at distance 0.3 m Dwight equation 13 under-predicts the
resistance by nearly 11%.The reasons for this difference include:
o Shape of the anode is not long and thin (Dwight assumed the anode is a long thin
cylinder)
o The anode axis is not perpendicular to the insulated surface (Dwight assumed the
cylinder has axis perpendicular to the ground surface)
• For distance greater than 0.67 m, the resistance falls rapidly at first, so that at 2.0 m the
resistance is about 0.18 Ohms.
• The resistance reduces slowly at separation distance above 10 m, until at 60 m it is close to that
obtained in example 5 (which was for a single anode well away from any surface).

Figure 8: Variation of Resistance of a Single Cuboid Anode with distance from a Perfectly
Coated Surface
Figure 9: Detail of Variation of Resistance of a Single Cuboid Anode with distance from a
Perfectly Coated Surface

Figure 10: Contours of Normal Current Density on the Anode at Various Separation Distances
Example 7: Three Cuboid Anodes Arranged in an Equilateral Triangle at Various Distances
Apart

The first sled-like example involves three anode of the type shown in Figure 5, arranged in an
equilateral triangular shape as shown in Figure 11. The side length of the triangle was varied to
determine influence of anode spacing on resistance to ground of the sled. In this case the sled is
positioned well away from any structure, seabed or sea surface.

Dwight used the approximate method to study an equilateral triangular arrangement of ground rods,
and published the resulting interference ratio graphically as figure 4 in reference 1. Data extracted from
his graph is reproduced in this paper as the green curve in Figure 13. In this figure we show simulation
results as green squares.

For the case with anode spacing 0.5 m, the interference ratio is 0.49, which means that the anode
group will deliver 49% of the current which would be delivered by the 3 anodes if they were well
separated. The reason for the significant reduction can be seen in the contour plot (Figure 12) showing
distribution of normal current density over the surfaces of the anodes. The anode faces close to other
anodes show the smallest current density.

Discussion of example 7.
From Figure 13 it is clear that the data published by Dwight showing interference between three ground
rods (arranged in an equilateral triangle) can with restrictions be applied to anodes arranged on a sled
as shown in Figure 11. At spacing 1.524 m (i.e., 5 ft) the data published by Dwight is about 8% too
high, probably as a result of the difference between the anode shape (Figure 11) and the ground rod
shape assumed by Dwight (roughly 3.0 m long, 20 mm diameter). If the Dwight data were to be
extrapolated to smaller spacing this error increases, for example at spacing 0.5 m (1.64 ft) the error
approaches 30%, i.e. the Dwight data exaggerates the sled current by 30%.

Figure 11: Views of the Sled used in Example 7, which has Three Anodes Arranged in an
Equilateral Triangle
 Figure 12: Contours of Normal Current Density for the Case where the Triangle Side Length is
0.5 m. Dark Blue is Smallest Current Density, Red is Biggest Current Density.

Figure 13: Solid Curves Reproduced from Figure 4 in Dwight. The Solid Green Line is Data for
Three Long Thin Ground Rods Arranged in an Equilateral Triangle. The Solid Red Line is Data
for Two Long Thin Ground Rods. The Solid Blue Line is Data for Four Long Thin Ground Rods
Arranged in a Square. Solid Green Squares are Simulation Results for Example 7. The “Ratio” is
the Interference Ratio Defined in Equation 7 Above.
Example 8: Seven Cylindrical Anodes Arranged on a Sled

The sled in this example has seven cylindrical anodes of length 2.1 m, diameter 0.25 m, spaced at 0.6
m between center lines (vertically and horizontally).

The only method available to find resistance to ground for this sled is simulation, which shows that the
interference ratio is 0.323, which means that the sled will deliver only 32% of the current which would
be delivered by the 7 anodes if they were well separated.

The contour plot of normal current density in Figure 14 clearly shows the interference which causes
large parts of the anode surfaces to have very low current density.

Figure 14: Contours of Current Density on Anode Surfaces. Blue = Low Current Density, Red =
High Current Density.

Example 9: 23 Cuboid Anodes Arranged on a Sled

In the final example we consider a sled with anodes in 3 layers. There are 8 anodes in the bottom and
top layers, and 7 anodes in the middle layer. The anodes (dimensioned as shown in Figure 5) are
spaced at 0.9 m between centers horizontally, with 0.9 m between layers (see Figure 15).

The contour plot of normal current density in Figure 16 clearly shows the interference which causes
large parts of the anode surfaces to have very low current density.

Using simulation to find resistance to ground for this sled, we find that the interference ratio is 0.192,
which means that the sled will deliver only 19.2% of the current which would be delivered by the 23
anodes if they were well separated.

By extrapolation of the curves in Figure 4 the interference ratio was estimated to be 0.145 for 23
anodes at 0.9 m spacing (2.95 ft). The simulation results show that this figure underestimates the value
of the interference ratio by nearly 25%, i.e. it will under-predict the sled current output by 25%.
Figure 15: Side View of the 3 Layer Sled used in Example 9, which has 8 Anodes on Top and
Bottom Layers and 7 Anodes in the Middle Layer

Figure 16: Contours of Normal Current Density on Surfaces of the Anodes on one Quarter of the
Sled (Result is Symmetric). Blue = Low Current Density, Red = High Current Density.

CONCLUSION

By comparison with analytic equations using examples for which the analytics were derived it has been
possible to show that simulation results obtained in this work are very accurate.

Since the simulation method can be applied with equal accuracy to any geometry, it can be concluded
that simulation can be used to measure the accuracy of approximate equations.

Examples have been given investigating the accuracy of Dwight and other equations when applied both
to geometry for which the approximate equations were derived and to anode shapes frequently used in
marine cathodic protection.

It has been shown that depending on separation distance of an anode from an insulating surface,
accuracy of the approximate equations for single anodes may be significantly in error.
Examples have been given investigating the accuracy achieved when approximate interference data,
derived for groups of ground rods, is applied to anode shapes frequently used in marine cathodic
protection

It has been shown that for large separation distances (between anodes in a group) the approximate
data is reasonably accurate, but for smaller separation distances the accuracy degrades significantly.

The term “interference ratio” has been used to describe the effect on output of interference between
anodes on a sled.

Finally, simulation has been applied to two examples of anode sleds, for which the interference ratio
was found to be 0.32 in one case and 0.19 in another.

REFERENCES

1. H.B. Dwight, “Calculation of Resistances to Ground,” Electrical Engineering, December 1936: p.

1319.

2. E. Hallen, “Losung zweir Potentialprobleme der Elektrostatik” Arkiv fur Matematik, Astronomi och
Fysik, v 21A, No 22, 1929 Stockholm

3. Formula for “Long slender stand-off” anode in Table 10-7 in document “Recommended Practice
DNV-RP-B401” January 2005 (Amended October 2006).

4. Formula for “Long flush mounted” anode in Table 10-7 in document “Recommended Practice DNV-
RP-B401” January 2005 (Amended October 2006).