| DOI: 10.3933/APPLRHEOL-28-54515 | WWW.APPLIEDRHEOLOGY.

ORG

Rheological Properties of Drilling Fluids: Use of Dimensionless

Shear Rates in Herschel-Bulkley and Power-law Models

Arild Saasen*1 and Jan David Ytrehus2

1 Department of Energy and Petroleum Engineering, University of Stavanger, 4036 Stavanger, Norway
2 SINTEF Industry, SINTEF, 7465 Trondheim, Norway

*Corresponding author: [email protected]

Received: 6.7.2018, Final version: 15.8.2018

Abstract:
An approach of Nelson and Ewoldt [1] to create a viscosity model of the Herschel-Bulkley type in order to use only parameters
with the potential of containing fluid information has been extended to be applied to drilling fluids using current industry
standard procedures. The commonly used Herschel-Bulkley consistency parameter k is found inadequate in describing fluid
properties properly as it has a unit dependent on n. Hence, the model is not optimum for digitalisation. The Herschel-Bulkley
model is re-written and base its parameters directly on the yield stress and the additional or surplus shear stress at a pre-deter-
mined shear rate relevant for the flow situation to be considered. This approach is also applicable for Power-Law models.

Key words:
Herschel Bulkley model, power-law model, viscosity parameters, drilling fluids, Nelson and Ewoldt approach

1 INTRODUCTION Several models are used to describe complex fluids

like drilling fluids. The range of these models include
The current practice, independent on type of industry, simple two-parameter models like the Bingham model
is to use digitalised models for industrial processes to complex models trying to encapsulate structure
whenever possible. Hence, different drilling processes build-up and disruption like the Quemada model [7, 8].
can be controlled by use of simple computer applica- Also viscoelastic properties can be important [5, 9, 10].
tions or, in many cases, also mobile phone apps. To be The simplest model that describes the flow curve with
applicable, these models must be reasonably accurate. reasonable accuracy seems to be the Herschel-Bulkley
This argument is also valid for drilling fluid viscosity model, named after Herschel and Bulkley [11], who de-
measurements. Current drilling practice rely on stan- scribed how such a flow curve should behave. In the
dards like API [2] or ISO [3, 4]. Even though the number Herschel-Bulkley model the shear stress is related to a
of measurement points may be limited, these stan- yield stress τy, a consistency factor k, and the shear rate
dards base their viscosity models on measurements g· by the use of Equation 1.
conducted at a wide range of shear rates. Earlier, the
viscosity models were based on viscosity measure-
ments at shear rates of 511 and 1022 1/s to create their
viscosity data. These shear rates are far too large to rep- (1)
resent practical drilling operations. It is recommended
in the current standards to use a least square fit of all The yield stress is a property arising from the composi-
shear stress measurements, using their affiliated shear tion of the drilling fluid. This value will change depen-
rates, to increase the accuracy of the viscosity models. dent on several parameters; for example the number
However, with the exception of the flow around the of particles of a certain size in the fluid. The unit of con-
Bottom Hole Assembly (BHA), shear rates in excess of sistency factor k is dependent on the curvature expo-
250 1/s are seldom experienced in the field [5, 6]. There- nent n, thus, k = k(n). The consequence of this is that the
fore, an improved model accuracy will be obtained if parameter cannot be determined directly from the fluid
the least square fit is conducted only for the relevant measurements and must be identified through alge-
shear rates of the drilling fluid flow situation. braic operations and that it cannot contain information

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 1 |

 limitations appearing when using the k and n approach.
This model will be explained in the next section. Still,
this model is not practical for describing drilling fluids
in accordance with the API or ISO specifications. There-
fore, based on this model a set of parameters is devel-
oped to modify the Nelson and Ewoldt parameters to
be applicable to drilling fluid engineering. These para-
meters should be applicable to add other types of infor-
Figure 1: The viscosity flow curves of a fluid presented with
mation like effects of vibration on the drilling fluids,
different k and n with yield stress is 0.2 Pa (red curve:
k = 0.0548 Pasn and n = 0.8269, blue curve: k = 0.0229 Pasn which so far not have been properly modelled. The ef-
and n = 0.9806). fects of vibration on drilling fluid flow curves have been
described, but not the effect on other parameters than
the yield stress [14].
Least square fit Yield stress, τy (Pa) Consistency, k (Pa.sn) n (dimensionless)
shear rate range
(1/s)
0 - 1000 0.2 0.0548 0.8269
0 - 300 0.2 0.0229 0.9806 2 USE OF NELSON AND EWOLDT’S PARA-
METERS IN THE HERSCHEL-BULKLEY MODEL
Table 1: Herschel-Bulkley parameters used in Figure 1.
Nelson and Ewoldt [1] found that by applying the con-
about physical dependencies for the fluid. An example sistency parameter k, which was dependent on n, it was
of relatively similar flow curves for different combina- not possible to compare different materials. Hence, pa-
tions of k and n is shown in Figure 1. Actually, both curves rameters are needed that are more universal than the
in Figure 1 are adopted from two approximations pre- traditional Herschel-Bulkley parameters. Nelson and
sented by Ytrehus et al. [12] for a field applied drilling Ewoldt [1] developed an alternative parameter to the
fluid. The viscosity measurements were conducted on consistency parameter k. Their first step was to deter-
an Anton Paar MCR102 Rheometer. Both curves repre- mine the yield stress from the viscosity flow curve. The
sent least square fit of rheometer data after the deter- yield stress is a fluid structural parameter. Then, they
mination of the yield stress. The curve with n = 0.8269 selected the shear stress and affiliated shear rate at
are constructed on the basis of using all measurement which the shear stress is twice the yield stress, named
values up to a shear rate of 1000 1/s, and the other by the critical shear rate.
using only the measurement values up to a shear rate
of 300 1/s to cover the laboratory experiment shear
rates. The viscosity model of the latter approximation (2)
was found to reproduce experimental pressure loss da-
ta more closely when the annular pressure loss model By introduction of this relation into the original Her-
by Founargiotakis et al. [13] for Herschel-Bulkley fluids schel-Bulkley equation (Equation 1) they obtained a
was used. Herschel Bulkley equation where all parameters are in-
Both approximations for the flow curve shown in dependent.
Figure 1 represents the measurements relatively well
even though the pair of parameters n and k are signif-
icantly different for these two curves. These parame-
ters are tabulated in Table 1. It is shown that the numer-
ical values of the consistency parameter developed by
fitting measurements values from the shear rate range (3)
up to 300 1/s is less than half of the value obtained if
all the measurement values up to 1000 1/s were used Equation 3 is a presentation of a Herschel-Bulkley fluid
in the fit. This is an example of the fact that k cannot in a form where all the parameters τy, g· c, and n can be
be used alone as a fluid property parameter. Its numer- treated separately. Furthermore, the shear rate enters
ical value will always be dependent on the index n. a dimensionless form in the equation. When measured
Hence, the meaning of tabulating the parameter k for properly with small enough increments, these parame-
other perspectives than reproducing numerical calcu- ters can all be determined directly from measurements.
lations should be questioned. Measurements of the field applied drilling fluid shown
Nelson and Ewoldt [1] presented a modified Her- in Figure 1 show that this drilling fluid had a very low
schel-Bulkley model with the scope of overcoming the yield stress. The yield stress of some drilling fluids are

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 2 |

too low to be determined by current oil well drilling the shear stress is set and the shear rate is measured,
standard procedures. These drilling fluids can be suffi- the Nelson and Ewoldt [1] model is obtained by setting
ciently well described using the power-law model that τs = τy and the corresponding g· c = g· s. The curvature ex-
do not exhibit any yield stresses. Furthermore, deter- ponent can be found for example by using Equation 6.
mination of particular shear stresses from use of con- In this case g· x is a selected shear rate where τx is mea-
ventional viscometers used in accordance with API/ISO sured. This shear rate should in principle be within the
specifications is not practical as these standards specify relevant shear rate range for the flow problem to be
measurements at a very limited number of shear rates. evaluated.
Also, the accuracy of these conventional VG meter mea-
surements at low shear rates can be questioned. There-
fore, it is not practical to use the parameters suggested
by Nelson and Ewoldt [1] for drilling fluids. However, it
is straightforward to expand their parameters to be (6)
used in drilling fluids.
The curvature exponent n as presented in Equation 6
will change if different shear rates are used to deter-
3 EXTENSION OF NELSON AND EWOLDT’S mine either n or τs. Therefore, the shear rates used to
PARAMETERS TO BE USED IN HERSCHEL- calculate n and τs must be specified. In principle it is now
BULKLEY MODELS FOR DRILLING FLUIDS possible to tabulate information or pressure and tem-
perature to n and τs measured at these specified shear
By selecting a relevant shear rate for the flow that shall rates. In general, it will not be possible to correlate data
be described using the Herschel-Bulkley model, it is measured at a particular shear rate directly to n and τs
straightforward to expand Nelson and Ewoldt’s ap- measured at other shear rates. However, this approach
proach and use parameters that can more easily be used will allow the industry to compare fluids planned for
in digitalised models. The first item is to approximate use in different well sections as these section have typ-
the yield stress from the viscosity flow curve. The next ical maximum shear rates during drilling fluid circula-
item is to determine a surplus stress τs at a specified tion.
shear rate. As an example, a shear rate of 170.3 1/s will
be selected in the next section. This shear rate is a rel-
evant for many drilling operations. This shear rate is 4 APPLICATION EXAMPLE
equivalent to that obtained at 100 RPM on most con-
ventional viscometers currently described by API [2] or In the following example the viscosity of drilling fluid
ISO [3, 4]. At the same time it is not too far outside the laboratory samples made for field application was eval-
applicable shear rate range for drilling fluid circulation. uated. For the ease of understanding the example is
made from direct application of simple measurements
with field drilling fluids. Note that improved results
may be obtained if results are fitted to the model prior
to the analysis in the example as is described in the fol-
(4) lowing chapter. Measurement data were collected at
the API [2] and ISO [3, 4] specified viscometer rotation
where rates. These rotation rates and the corresponding shear
rates are tabulated in Table 2. The first step is to approx-
imate the yield stress from the flow curve. In the current
examples the yield stress is approximated following
(5) Zamora and Power [15] as:

For a power-law fluid, Equations 4 and 5 are still valid

by setting τy = 0. If this equation is reversed such that (7)

RPM 3 6 30 60 100 200 300 600

1/s 5.11 10.22 51.1 102.2 170.3 340.7 511 1022

Table 2: Conversion of VG meter RPM to shear rates (1/s).

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 3 |

Figure 2: Comparison of measurement values with model Figure 3: Comparison of measurement values with model
predictions for an oil based drilling fluid measured at 20°C predictions for an oil based drilling fluid measured at 20°C
(τy = 4.6 Pa, τs = 15.33 Pa, and g· s = 170.3 1/s). Solid line repre- (τy = 4.6 Pa, τs = 15.33 Pa, and g· s = 170.3 1/s). Solid line repre-
sents results calculating n = nls = 0.695 at g· = 50.11 1/s and sents results calculating n = nls = 0.695 at g· = 50.11 1/s and
the dotted line represents the results calculating n = nhs = the dotted line represents the results calculating n = nhs =
0.796 at g· = 1022 1/s. 0.796 at g· = 1022 1/s. The figure is an enlargement of the low
shear rate regime of Figure 2.

Figure 4: Comparison of measurement values with model Figure 5. Comparison of measurement values with model
predictions for a water based drilling fluid measured at 20°C predictions for a water based drilling fluid measured at 50°C
(τy = 5.11 Pa, τs = 18.4 Pa, and g· s = 170.3 1/s). Solid line repre- (τy = 2.56 Pa, τs = 15.33 Pa, and g· s = 170.3 1/s). Solid line repre-
sents results calculating n = nls = 0.674 at g· = 50.11 1/s and sents results calculating n = nls = 0.695 at g· = 50.11 1/s and
the dotted line represents the results calculating n = nhs = the dotted line represents the results calculating n = nhs =
0.717 at g· = 1022 1/s. 0.547 at g· = 1022 1/s.

In the examples the numerical subscripts refer to the g· = g· s = 170.3 1/s and the other one at higher shear rates.
particular rpm of the conventional viscometers de- The curvature exponents for these two cases are de-
scribed in API [2] and ISO [3, 4] procedures, albeit the fined as nls for the low shear exponent and nhs for the
fact that only metric units are used in the calculations. high shear exponent. In the current example the mea-
The next step is to measure the shear stress at 170.3 1/s. surement at 30 rpm (51.1 1/s) is used to produce nls and
This value was chosen as it represents a typical upper the measurement at 600 rpm (1022 1/s) is used to create
limit for a lot of practical annular flow cases [5, 6]. Then nls. Hence, these two values are calculated as:
the yield stress was subtracted to give the surplus shear
stress τs.

(8)

This parameter does no longer contain any dimension

dependent on the curvature exponent n. Finally, the (9)
curvature exponent n is determined. In the present
cases two possibilities were chosen. One should have In Figure 2 it is shown the shear stress as function of
the optimum accuracy at the shear rates less than shear rate for a comparison of measurement values

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 4 |

 Data for Figures 2 and 3 Data for Figure 4 Data for Figure 5
g& Measured nls nhs Measured nls nhs Measured nls nhs
5.11 5.62 5.94 5.54 6.13 6.84 6.60 4.09 3.90 4.80
10.22 6.64 6.77 6.22 7.15 7.88 7.56 5.62 4.73 8.84
51.1 11.2 11.2 10.5 13.3 13.3 12.9 9.20 9.20 10.5
102.2 15.3 15.4 14.8 17.9 18.2 17.9 12.3 13.3 14.1
170.3 19.9 19.9 19.9 23.5 23.5 23.5 17.9 17.9 17.9
340.7 30.7 29.4 31.2 34.7 34.5 35.3 23.0 27.4 25.0 5 COMMENTS TO THE
511 40.9 37.5 41.4 46.0 43.7 45.5 28.6 35.4 30.5
1022 68.5 57.8 68.5 71.5 66.6 71.5 43.4 55.8 43.4
RESULTS
ty 4.60 4.60 5.11 5.11 2.56 2.56
ts 15.3 15.3 18.4 18.4 15.3 15.3
Traditionally, the low shear rate ex-
nls 0.695 0.674 0.695 periments using manual equipment
nhs 0.796 0.717 0.547
may have large uncertainties. That
Table 3: Measured and modelled data used in Figures 2 – 5. Numbers in bold face will question the measured yield
italics are exact values used to calculate τs and n. Unit for all data is Pa. stress. However, by use of more accu-
rate instruments, the determination
of the yield stress can be improved.
with model predictions for an oil based drilling fluid. All current measurements using k and n may be useful
The yield stress was τY = 4.6 Pa, the surplus shear stress in the proposed model if the k value is transferred to a
τs = 15.33Pa at the pre-determined shear rate g· s = 170.3 surplus shear stress value. It is also a benefit if the mod-
1/s. The solid line represents results that are more ac- els are optimised by curve fitting within the relevant
curate at lower shear rates using n = nls = 0.695. The dot- shear rate range. Such curve fit models will normally re-
ted line the shows the results that are more accurate at duce some of the uncertainties introduced by using the
higher shear rates with the exponent n = nhs = 0.796. To standard measurement procedures. The surplus shear
compare these results at the lower shear rates it is prac- stress can be calculated as:
tical to evaluate Figure 3 which represents a low shear
rate magnification of Figure 2.
It is shown in Figure 3 that with the selected shear
rates, all models have identical yield stress values and (10)
equal values at the shear rate of 170.3 1/s. At the shear rate
of 51.1 1/s the curve with n = nls has an identical value as If Equation 10 is inserted into Equation 4, the original
the experiments. This specific measurement was selected Herschel-Bulkley equation is resumed. The current
to determine the value of the exponent n. Similarly, it is practice to describe the drilling fluids with k and n hin-
observed in Figure 2 that with n = nhs an identical value as ders optimum digitalisation process within the drilling
the experiments is obtained at the shear rate 1022 1/s. industry. By changing to evaluate the yield stress, the
While working with drilling fluids it is easy to get the im- surplus stress and the curvature exponent at a relevant
pression that a Herschel-Bulkley modelled flow curve de- shear rate, the parameters can easier be related to tem-
velop lower values if the model is based on the lower shear perature, pressure, chemical content, and particle ad-
rate values like what is shown in Figure 3. This is, however, dition. Hence, this approach will give the field engineer
not generally true. For the curve shown in Figure 3, the pre- a better understanding of the function of the viscosity
dicted shear rate values for the curve with n = nhs give a model.
lower shear stress value at the lower shear rates with n =
nls. The flow curves shown in Figure 4 and Figure 5 illustrate
a different behavior. Because the values at the shear rate 6 CONCLUSIONS
at 170.3 1/s is identical (see Table 3), the curve with the low-
est n value will show the lowest shear stress at lower shear A viscosity model of the Herschel-Bulkley type where
rates and highest at higher shear rates. However, the the shear rate is made dimensionless by selecting a
curves may have its pronounced curvature at different characteristic shear rate for the flow has been present-
shear rates. First of all, as was shown already in Figure 1, ed. An example is presented where this approach is
two curves with different shear stresses may give approx- used on drilling fluids. The Herschel-Bulkley parame-
imately the same results. This is also shown in Figure 4 for ters may then have the potential of containing fluid in-
a set of measurements on a water based drilling fluid at formation and be compared with other fluids. The Her-
20 °C. When the measurements of this water based schel-Bulkley consistency parameter k is found inade-
drilling fluid were conducted at a temperature of 50 °C, quate in describing properties in a simple way as it has
the situation is different. First the yield stress is reduced a unit dependent on n. Hence, the model is not opti-
to the half. The surplus stress at 170.3 1/s is also reduced. mum for digitalisation. The Herschel-Bulkley model
However, the shear rate range with the highest curvature could be re-written as τ = τy + τs (g· /g· s)n where τy is the
has been altered. The most accurate results at lower shear yield stress and τs = τ - τy at the pre-determined g· = g· s.
rates are now found using n = nls = 0.695, which is larger The proposed method is equally good for the descrip-
than n = nhs = 0.547 that predicts more accurate results at tion of power-law models; simply by setting the yield
higher shear rates. This is illustrated in Figure 5. stress equal to zero.

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 5 |

 [7] Qemada, D: Rheological modelling of complex fluids I.
ACKNOWLEDGEMENTS The concept of effective volume fraction revisited, Eur.
Phys. J., Appl. Phys. 1 (1998) 119 – 127.
The authors thank Professor Randy Ewoldt at the Uni- [8] Baldino, S, Osgouei RE, Ozbayoglu E, Miska SZ, May R:
versity of Illinois for fruitful discussions on this subject Quemada model approach to oil or synthetic oil based
during the Nordic Rheology Conference in Trondheim, drilling fluids rheological modelling, J. Petrol. Sci. Eng.
Norway (2018). 163 (2017) 27 – 36.
[9] Bizhani M, Kuru E: Particle removal from sandbed de-
posits in horizontal annuli using viscoelastic fluids, SPE
J. 23 (2018) 189443-PA.
REFERENCES [10] Bizhani M, Kuru E: Critical review of mechanistic and em-
[1] Nelson AZ, Ewoldt RH: Design of yield stress fluids: A rhe- pirical (semimechanistic) models for particle removal
ology-to-structure inverse problem, Soft Matter 13 (2017) from sandbed deposits in horizontal annuli with water,
7578 – 7594. SPE J. 23 (2018) 187948-PA.
[2] American Petroleum Institute: API Recommended Prac- [11] Herschel WH, Bulkley R: Konsistenzmessungen von
tice 13B2, Washington D.C. (2014). Gummibenzöllösungen, Kolloid Z. 39 (1926) 291 – 300.
[3] International Organization for Standardization: Petrole- [12] Ytrehus JD, Lund B, Taghipour A, Kosberg B, Carazza L,
um and natural gas industries – Field testing of drilling Gyland KR, Saasen A: Cuttings bed removal in deviated
fluids. Part 1: Water-based fluids, Report ISO 10414-1, wells, ASME 37th Int. Conf. Ocean, Offshore and Arctic
Geneva (2008). Eng., Madrid (2018).
[4] International Organization for Standardization: Petrole- [13] Founargiotakis K, Kelessidis VC, Maglione R: Laminar,
um and natural gas industries – Field testing of drilling transitional and turbulent flow of Herschel-Bulkley flu-
fluids. Part 1: Oil-based fluids, Report ISO 10414-2, Geneva ids in concentric annulus, Canad. J. Chem. Eng. 86 (2008)
(2011). 676-683.
[5] Werner B, Myrseth V, Saasen A: Viscoelastic properties [14] Saasen A, Hodne H: The influence of vibrations on
of drilling fluids and their influence on cuttings trans- drilling fluid rheological properties and the consequence
port, J. Petrol. Sci. Eng. 156 (2017) 845 – 851. for solids control, Appl. Rheol. 26 (2016) 25349.
[6] Sayindla S, Lund B, Ytrehus JD, Saasen A: Hole-cleaning [15] Zamora M, Power D: Making a case for AADE hydraulics
performance comparison of oil-based and water-based and the unified rheological model, AADE Techn. Conf.,
drilling fluids, J. Petrol. Sci. Eng. 158 (2017) 49 – 57. Houston (2002).

© Appl. Rheol. 28 (2018) 54515 | DOI: 10.3933/ApplRheol-28-54515 | 6 |