SPE 95088

A Complete Theory of Scale-Inhibitor Transport and Adsorption/Desorption in

Squeeze Treatments
K.S. Sorbie, Heriot-Watt U., and R.D. Gdanski, Halliburton

Copyright 2005, Society of Petroleum Engineers Inc.

wellbore in the produced brine over an extended period.
This paper was prepared for presentation at the SPE International Symposium on Oilfield However, the mechanism of this retention is open to
Scale held in Aberdeen, United Kingdom, 11–12 May 2005.
interpretation and is known to depend on various conditions
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
including the SI type, temperature, pH, [Ca2+], nature of
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to mineral substrate, etc. These conditions are either deliberately
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the SPE, their officers, or members. Electronic reproduction, distribution, or storage imposed by the treatment design or they exist in the reservoir.
of any part of this paper for commercial purposes without the written consent of the Society of
Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract
Broadly speaking, it is widely accepted that the two principal
of not more than 300 words; illustrations may not be copied. The abstract must contain mechanisms of SI retention within a reservoir formation are
conspicuous acknowledgment of where and by whom the paper was presented. Write
Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. due to adsorption and precipitation. In the general case, a
combination of these mechanisms may be involved and this
can be described mathematically. Here we focus on where
Abstract
adsorption is the main retention mechanism.
Software tools for designing downhole scale inhibitor (SI)
squeeze treatments are widely available and have been
A number of papers have appeared describing the modelling
successfully applied in thousands of wells worldwide. Many
of SI transport and retention during flow through porous
SPE papers have been published describing such squeeze
media [1-6]. In particular, alternative approaches for
design case studies from the conceptual modelled design, to
modelling adsorption squeeze processes have been developed
implementation and post treatment analysis. These models are
by the authors [3, 6] and some confusion has arisen as to how
based on the fundamental transport equations for the SI in the
these are related. In this paper, we will clarify the similarities
near well formation. These equations must also include a
and differences between these approaches and point to
model describing the SI/rock interaction, regardless of whether
advantages and disadvantages of each. The main aspects of the
it occurs by an adsorption/desorption or by a precipitation
models proposed relate to (a) the mathematical structure of
(phase separation) mechanism. A number of papers have
the transport/retention equations describing the SI within the
appeared in the literature on these fundamental equations and
porous medium; and (b) the surface chemistry assumptions
on the analytical and numerical models based on them.
inherent in the various models that are used to describe the
specific details of the SI/rock retention mechanism. Thus, our
In this paper, we present a re-evaluation of the equations that
analysis and comparison will be carried out in the context of
have been proposed to model SI transport and adsorption in
the mathematics and the description of the adsorption process.
porous media. We have analysed the various approaches in
terms of two basic aspects: (a) the mathematical structure of
It will be shown that a unified mathematical model emerges
the various equations used to describe transport; and (b) the
where certain choices or assumptions must be made on the
surface chemistry assumptions and models used to describe
form and kinetics of the actual adsorption model. Depending
the SI/rock retention mechanism, particularly by adsorption.
on these choices/assumptions, the HW (Heriot-Watt) or
We specifically focus on comparing and reconciling our own
Halliburton models are recovered. The differences are then
(Heriot-Watt U. and Halliburton) respective approaches,
those of detail – although this detail may be very important in
which have been developed over the last few years.
specific circumstances. Although we focus on adsorption SI
treatments in this work, precipitation type models also fit into
Background and Introduction
a similar framework with a modified retention model [7, 8].
Scale inhibitor (SI) squeeze treatments have been carried out
in many reservoirs as a measure for controlling/preventing the The Scale Inhibitor Transport/Retention Equations
formation of oilfield mineral scales such as barium sulphate
and calcium carbonate. Such treatments are routine in both Generalised convection-dispersion equation: The generalised
onshore and offshore situations and great expertise exists in convection-dispersion equation describing SI transport and
the service sector and the operator companies in performing kinetic (i.e. non-equilibrium) adsorption is as follows:
these treatments quite successfully. A key feature of the ⎛ ∂c ⎞ ⎛ ∂ 2c ⎞ ⎛ ∂c ⎞ ρ ⎛ ∂Γ ⎞
⎜ ⎟ = D⎜ 2 ⎟ − v⎜ ⎟ − ⎜ ⎟
process is the nature of the SI/rock interaction which retains (1)
⎝ ∂t ⎠ ⎝ ∂x ⎠ ⎝ ∂x ⎠ φ ⎝ ∂ t ⎠
the SI within the formation, allowing it to return to the
2 SPE 95088

where c = [SI] (g/L or mg/cm3), D = dispersion (cm2/s), fluid ⎛ ρ ⎛ ∂Γ ⎞ ⎞ ⎛ ∂c ⎞ ⎛ ∂ 2c ⎞ ⎛ ∂c ⎞ (4)

velocity, v = Q/(Aφ) (cm/s), ρ = density (g/cm3) and φ = ⎜ 1+ ⎜ ⎟⎟⎜ ⎟ = D ⎜ 2 ⎟ − v⎜ ⎟
⎝ φ ⎝ ∂c ⎠ ⎠ ⎝ ∂t ⎠ ⎝ ∂x ⎠ ⎝ ∂x ⎠
porosity (dimensionless). Γ(c ) is the level of adsorption in
units of mg/g in Eq. 1, but different units are used by various The non-linear term which now multiplies the LHS (left hand
authors (depending on units of c) leading to different forms for side) of Eq. 4, is very important since it provides a
the adsorption term in Eq. 1, as listed is Table 1. “retardation factor” that explains the long tail in SI return
curves at low concentration levels, as explained in several
Table 1: Units of adsorption and how these relate to the form references [3, 9 - 11]. This can be seen by ignoring dispersion
of the adsorption rate term in the transport equation, Eq. 1. (D = 0), which is quite reasonable in the long tail, and
rearranging Eq. 4 to obtain:
Design- Units of SI Mass SI per Form of the rate
ation adsorption rock volume, VT term in Eq. 1 ⎛ ∂c ⎞ v ⎛ ∂c ⎞ ⎛ ∂c ⎞ (5)
⎜ ⎟= − ⎜ ⎟ = −vc ⎜ ⎟
[ref.] (cm3) where c => ⎝ ∂t ⎠ ⎛ ρ ⎛ ∂Γ ⎞ ⎞ ⎝ ∂x ⎠ ⎝ ∂x ⎠
(mg) g/L ⇔ mg/cm3 ⎜1+ φ ⎜ ∂c ⎟ ⎟
⎝ ⎝ ⎠ ⎠
ΓA mg SI Γ A ρVT or* ρ ⎛ ∂Γ ⎞
− ⎜ A⎟
[3] g rock Γ A (1 − φ ) ρrgVT φ ⎝ ∂t ⎠ where v c is the velocity of concentration level c (in the
Lagrangian sense) and we see that the retardation factor is
ΓB mg SI Γ BVT φ −
1 ⎛ ∂Γ B ⎞ bigger if the isotherm is very steep i.e. ( ∂Γ ∂c ) >> 1 . Because
⎜ ⎟
L of pore space 1000 ⎝ ∂t ⎠
1000 the shape of the adsorption isotherm is much steeper at low
concentrations, then the velocity v c for these lower
ΓC mg SI ΓCVT φ ⎛ ∂Γ ⎞
−⎜ C ⎟
cm3 of pore space ⎝ ∂t ⎠
concentrations (in a core flood or in the field) is very low,
[6]
compared with the transit time through the core. For the high
SI concentrations where c = co (the injected level), then this
ΓD mg SI Γ DVT 1 ⎛ ∂Γ D ⎞
− retardation factor is ≈1 and the corresponding velocity is about
⎜ ⎟
L of rock 1000 1000φ ⎝ ∂t ⎠ the same as the fluid velocity, i.e. v c = c = v . Hence, the higher
ΓE Γ E (1 − φ ) VT (1 − φ ) ⎛ ∂Γ E ⎞
0
mg SI
− SI concentration levels travel at about the fluid velocity and
[2]
L of rock grain 1000 φ ⎜⎝ ∂t ⎟⎠ may be returned to the well relatively rapidly, which accounts
for the large early-time “spike” that is frequently observed in
* where ρ = rock density (grain and pore space); ρ rg = rock grain field squeeze treatments [9 – 11].
3
density; both g/cm
Kinetic (Non-equilibrium) Adsorption and the Form of the
The simplest form of the adsorption rate term is obtained when Equilibrium Adsorption Isotherm
c and Γ(c ) are in the same units – say, mg/L of pore space – Background: Let us first revisit the development of classical
since this gives in a bulk, single cell system (e.g. a static isotherms. It is sometimes implied that a single “adsorption
beaker SI adsorption test) that: isotherm”, Γ(c), is not adequate to describe both adsorption
⎛ ∂c ⎞ ⎛ ∂Γ ⎞ and desorption processes, since these may occur at very
⎜ ⎟ = −⎜ ⎟ (2) different rates. This can be analysed thermodynamically. It is
⎝ ∂t ⎠ ⎝ ∂t ⎠ certainly possible that, in a field (or laboratory) squeeze
treatment, the return flow conditions of temperature, salinity,
Eq. 1 is a mass conservation equation and makes no reference pH, etc., may be different and hence a different adsorption
to the mechanism of adsorption/retention, the form of the isotherm should apply. This complex behaviour is modelled
adsorption isotherm, Γ(c ) , or to the kinetic rate law used for using the phenomenon of “coupled adsorption”, as described
( ∂Γ ∂t ) . These are matters which only experiment can settle previously [10, 12]. However, for the same conditions (pH, T,
and, if a certain adsorption isotherm or rate law is assumed, brine composition, etc.), one isotherm applies relating
then it must subsequently be tested against experiment. specifically to the condition where dynamic adsorption rate is
equal to the dynamic rate of desorption. The actual
Equilibrium adsorption and the adsorption isotherm: Before mathematical form of Γ(c) emerges from the assumed rate
considering either the form of the adsorption isotherm, Γ(c) , laws for adsorption/desorption, as was demonstrated by
Langmuir and others. For example, consider a single mobile
or the rate law in Eq. 1, we first note that if the system is at
phase with c = [SI] and an adsorbing substrate where the
equilibrium, then the appropriate equilibrium adsorbed level of
adsorption level is Γ in appropriate units, under fixed fluid
SI is reached instantaneously. Therefore, Γ depends only on c,
conditions of pH, T, etc. The adsorption and desorption rates
and so the adsorption term by the chain rule becomes:
are denoted Ra and Rd, respectively. Taking the assumed
ρ ⎛ ∂Γ ⎞ ρ ⎛ ∂Γ ⎞ ⎛ ∂c ⎞ (3)
− ⎜ ⎟ = − ⎜ ⎟⎜ ⎟ forms for these rates as:
φ ⎝ ∂t ⎠ φ ⎝ ∂c ⎠ ⎝ ∂t ⎠ Ra = ka .c n (adsorption) (6)
and
This term can be introduced into Eq. 1 and rearranged to
obtain the equilibrium transport equation for the SI which is: Rd = kd .Γ(c) (desorption) (7)
SPE 95088 3

where ka and kd are the adsorption and desorption rate expressions for all 3 quantities, Rd , Ra and Γ eq (c) . Before
constants, respectively, and the constant n is in the range 0 < n discussing the actual form of the adsorption isotherm, we first
< 1. Recognizing that at equilibrium Ra = Rd , it follows that consider the kinetic rate law and what implications this has.
the adsorption isotherm that arises is:
Kinetic rate laws: In the light of the above paragraph, we now
Γ (c ) = α .c n (8)
consider the form of the kinetic adsorption rate expression
where we acknowledge that the timescale for adsorption may
where this is the Freundlich isotherm and having the constant be comparable with that of the flow times. For example, it is
α = ( ka / kd ) . A linear form of Γ(c) arises if n = 1 because of known experimentally that the time to reach full equilibrium
the assumed linear rate laws for adsorption and desorption. adsorption for a phosphonate onto quartz sand is of order 1 – 3
Note that very different rates of adsorption and desorption are hours [13]. If the transit time through a core is significantly
not only admitted but are required. This makes physical sense less than this, the system will not reach equilibrium. Note the
since the slower the desorption (the smaller is kd ) then the subtlety that this transit time through the core is concentration
higher is α, which is the slope of the adsorption isotherm. dependent, from the discussion about vc above.

Alternatively, a more complex pair of adsorption/desorption Consider the form of the kinetic adsorption term, ( ∂Γ ∂t ) , in
rate laws may be assumed. For example, this may be done to Eq. 1. As noted, if we make the Freundlich type assumptions
model the fact that a rock substrate may have a maximum (Eqs. 6 and 7), then the rate law is determined a priori as [2]:
adsorption capacity, Γmax, where no more SI can be adsorbed
⎛ ∂Γ ⎞
⎟ = Ra − Rd = k a .c − k d .Γ (c )
n (13)
even if the mobile phase concentration, c, is increased further. ⎜
Firstly, we define the fractional coverage of the surface, f (also ⎝ ∂ t ⎠
denoted by Γi (c ) ), by adsorption of the SI, as follows:
Other expressions have been proposed for this rate law as
f =Γi (c) = Γ eq (c) (9) follows [3]:
Γ max ⎛ ∂Γ ⎞
⎜ ⎟= κ (Γ (c) − Γ (t ) ) (14)
⎝ ∂t ⎠
eq

where Γ eq (c) is the equilibrium adsorption level at a given

mobile phase SI concentration, c. Clearly, if Γ eq = Γ max , then where κ (s-1) is the adsorption rate constant, Γ eq (c) is the
physically no further net adsorption of SI can take place from underlying equilibrium adsorption isotherm, and Γ(t ) is the
the mobile aqueous phase onto the rock surface. This leads current level of adsorption at time, t. This expression is
to the following pair of adsorption/desorption rate expressions: probably the simplest that has the correct form, i.e. as
Ra = ka .c(1 − f ) (adsorption) (10) Γ → Γ eq (c) , then the rate of adsorption tends to zero. Also, if
and Γ > Γ eq (c ) , then ( ∂Γ ∂t ) is negative and vice versa.
Rd = kd . f (desorption) (11)
An alternative form of the desorption rate expression has been
Thus, at equilibrium ( Ra = Rd ) we obtain the expression: proposed by Gdanski and Funkhouser (G-F) [6] as follows
α .c.Γ max ⎛ dc ⎞ 2
Γ= (12) ⎜ ⎟ = kdes ⎡⎣ceq − c0 ⎤⎦ .Γ (15)
(1 + α .c ) ⎝ dt ⎠

which is exactly the form of the Langmuir adsorption where k des is the desorption rate constant, c0 is identified with
isotherm. Clearly, at low c, the isotherm is linear and as the local solution concentration, c, and hence varies with time.
c → ∞ then Γ → Γmax , as required. Hence, the form of the Clearly, as c → ceq then the rate of desorption will tend to zero
equilibrium adsorption isotherm emerges from the forms of 2
the adsorption and desorption rate laws. quite rapidly since it depends on ⎡⎣ ceq − c0 ⎤⎦ . Considering a
static system (Eq. 2 applies), then we may transform the G-F
Thus, we can conclude that behind any view of the adsorption/ expression in Eq. 15, as follows. Note that the quantities c and
desorption process (for fixed fluid conditions of pH, T, etc.) Γ in the G-F paper [6] are in units of mg/cm3 of fluid in the
there is a single underlying equilibrium adsorption isotherm, pore space. Then for initial conditions co and Γo, the initial
Γ eq (c) . Hence, if we wish to specify a desorption rate total mass per cm3 of fluid is mo = co + Γo (which in a static
expression, Rd , then along with a given adsorption rate system is fixed by mass conservation) and throughout the
dynamic process of desorption (adsorption) then c(t) + Γ(t) =
expression, Ra , this defines the form of, Γeq (c) .
mo, and likewise, ceq + Γeq = mo. Thus, the G-F desorption rate
Alternatively, if we define the desorption rate law and the equation can be transformed into an expression for the rate law
equilibrium isotherm, then doing so implicitly defines the for Γ(t), as follows:
adsorption rate law. Clearly, we cannot define independent
4 SPE 95088

⎛ d Γ(t ) ⎞
⎟ = − kdes ⎡⎣( m0 − Γ eq ) − ( m0 − Γ (t ) ) ⎤⎦ .Γ(t )
2 [14], they found that minerals studied can be broadly classified
⎜ into three groups, as follows: (i) strongly adsorbing minerals
⎝ dt ⎠
(16) (e.g. siderite); (ii) moderately adsorbing “silica-like” minerals
which gives: (e.g. silica and kaolinite); and (iii) weakly adsorbing “alumina-
like” minerals (e.g. illite, smectite, and alumina). Furthermore,
⎛ d Γ (t ) ⎞ 2
⎟ = − kdes ⎣⎡ Γ(t ) − Γ eq ⎦⎤ .Γ(t )
(17)
⎜ they noted that such static adsorption isotherms were useful in
⎝ dt ⎠ determining the magnitude of the kinetic effect on desorption
in linear flow tests. They also conjectured that siderite may be
Eq. 17 is referred to as the G-F adsorption rate expression. responsible for the long-term, low-level inhibitor-return
profiles often observed in squeeze treatments [14, 19].
Static solutions to rate laws: In fact, the static rate laws in
Eqs. 14 [3] and 17 [6] can both be integrated analytically using This leads to the second feature of the G-F approach where (i)
standard integrals. However, here we performed the a (sometimes modified) Langmuir adsorption isotherm is
integration numerically and the results have been compared assumed as the equilibrium isotherm, Γeq (c) , and (ii) that
for both rate expressions to compare the general character of
these two solutions. Results are shown in Fig. 1 for some different minerals in the rock matrix may have different
assumed kinetic parameters κ = 0.03 and kdes = 2, in arbitrary adsorption/desorption behaviour (characterized by their
1/α values). Indeed, following point (ii), it is reasonable that
units where the underlying Γeq (c) were the same Langmuir if the mineral composition of a reservoir rock is known, then
expressions. These parameters were chosen to get the early the fractions of the various relevant isotherms can be used to
time behaviour of c (i.e. the increase in mobile phase predict the outcome for a given system. This requires that a
concentration due to desorption) vs. time (t), to be roughly the “data base” of Langmuir isotherms for various SIs and
same. Note that these rate laws have very different longer different mineral substrates be built up, following the
time approaches to equilibrium. This is a significant practical approach given in refs. [3, 14].
difference between the two approaches since it implies that
rate effects will be intrinsically more important in the G-F In contrast, the approach of the HW group [3, 4, 9 – 11] has
approach. Which rate law is more correct is amenable to generally been uncommitted on the precise form of the SI/rock
experiment and, indeed, it may be that different rate laws for adsorption isotherm and this is discussed in detail below.
adsorption (Eq. 14, Eq 17, or possibly another form) are
required for the adsorption of different types of scale Discussion of Adsorption Results
inhibitors – phosphonate and polymer – on different types of Since the α parameter determines where the “50% loading” of
mineral substrate. Here, we simply note the important the isotherm lies (Fig. 2), it is interesting to examine how
difference for the present time. conventional experimental dynamic inhibitor/rock isotherms
plot on such diagrams. We wish to see if they show (1/ α)
Representing Equilibrium Inhibitor/Rock Adsorption values at very low values of c (≈10 ppm) and, if not, how this
A Langmuir adsorption isotherm for describing the can be interpreted. We approach this issue conceptually by
inhibitor/rock adsorption process is given in Eq. 12. Denoting plotting the inhibitor isotherms shown in Fig. 2 (normalised
the fractional coverage of the mineral surface by scale i (c ) vs. c ) as if they had a value of Γ
Γ = 1 mg/g (quite
i (c) , then:
inhibitor as Γ
max

reasonable for adsorbing scale/inhibitor sandstone systems).

i (c ) = Γ(c) = α .c
Γ (18) The corresponding isotherms as Γ (mg/g) vs. c (ppm) are
Γ max (1 + α .c ) shown in Fig. 4 where the isotherm for the highest value of
α = 1, is certainly very steeply rising and is not typical of the
where the coverage is seen to depend only on the single dynamic adsorption isotherms observed in the earlier HW
parameter, α, and it is shown above that, α = ( ka / kd ) . An work [3, 9 - 11]. For α = 1, the implication is that the scale
inhibitor is up to 95% of its maximum adsorption level ( Γ max
example of the fractional coverage vs. c is given for cases with
various α parameters in Fig. 2, where c is in ppm. A log scale = 1 mg/g) at ∼19 ppm – this is not observed in the total rock
was used to clearly visualise the lower concentration region. system in sandstones. However, it may occur in mineral clay
The concentration at 50% coverage is important in these plots separates as the case of siderite shows in Fig. 5 [14]. The data
as this can easily be shown to occur at c = 1/α. Fig. 3 shows for adsorption on siderite is not as clearly supportive of the
that experimental data for DETPMP (diethylenetriamine penta Langmuir form as that in Fig. 3 for kaolinite. However, it
methylene phosphonic acid) on kaolinite at pH 7 fits the does indicate high mineral coverage at very low [SI] thus
Langmuir curve quite well [14]. supporting the authors’ claim that it may contribute to the long
low returns in a squeeze treatment [14, 19].
Funkhouser and Gdanski [14] found that, in some systems,
static adsorption isotherm data do not follow the simple We now analyse the dynamic adsorption isotherms for
Langmuir isotherm. However, the modified Langmuir DETPMP found by the HW procedure for sandstone reservoir
expression [14] fitted their observations very well over a wide cores coming from a range of sources. In these experiments
variety of adsorption conditions. Following this approach the inhibitor was adsorbed from a low-pH mainstage. A non-
buffered overflush was flowed and the pH during desorption
SPE 95088 5

allowed to rise from about pH 2 or 3 up to about pH 6. In laws (Eqs. 14 and 17). This is done in Appendix A and it
some cases, precipitation of the DETPMP was induced within leads to various dimensionless numbers that are presented and
the core. As such, the effluent profiles included multiple interpreted in Table A1. Those referring to the balance
effects including a reduction in plateau adsorption, acid-base between adsorption rate (κ or k des ) and residence times (L/v)
neutralization, solubility limits and kinetics of desorption. The
for the two models are the Damköhler numbers, N Da1 and
dynamic adsorption isotherms derived from these floods are
shown in Fig. 6 over a wide SI concentration range N Da 2 . These are given by the following equations for the HW
(linear/linear scale). The same data are shown in Fig. 7 (a) on and G-F models, respectively:
a log concentration/linear adsorption scale and (b) in the lower ⎛κL ⎞
SI concentration region. The corresponding “validation” of N Da1 = ⎜ ⎟ (20)
these dynamic isotherms is confirmed by using SQUEEZE V ⎝ v ⎠
to model the SI return curves for each flood as shown for two and
examples (Floods C3 and C4) in Fig. 8 (All other floods – not ⎛ k .L.Γ 2max ⎞
shown – are matched equally well). The very good agreement N Da 2 = ⎜ des ⎟ (21)
shows that these isotherms are the ones that describe the ⎝ v ⎠
“effective processes” occurring in the cores. Under these These numbers are of very similar form, but the G-F
particular conditions, the process is probably adsorption, but expression contains an additional Γ 2max term due to the
the HW procedure does not specifically assume the form of quadratic non-linearity in the rate law in Eq. 17. In the limit of
the isotherm in the modelling process. The adsorption large Damköhler numbers, the adsorption is very fast and the
isotherms in Fig. 6 are generated as tables of numbers (c vs. Γ) system will be close to the equilibrium limit. For
without reference to either the detailed mechanism or the form “intermediate” values, the rates of convection and adsorption
of the isotherm, as described previously [3, 9, 11]. will be similar and at very low values, the system will be far
from equilibrium.
We note that the derived dynamic isotherms shown in Fig. 6
(and 7) may be expressed as fractional coverage, Γi (c) vs. c (in Rate effects – HW model: The effect of rate (deviation from
ppm) and these are compared with the model curves (Fig. 2) equilibrium) is shown for the HW model by varying the
for all 6 floods in Fig. 9(a). By separating out these curves, Damköhler number, N Da1 . The resulting SI effluent profiles
we observe that the dynamic Γi (c) vs. c curves either for the cases in Table 2 are shown in Figs. 11 – 13. These
approximately do or do not resemble the Langmuir isotherms calculations are based on the equilibrium isotherm for flood
as shown in Figs. 9(b) and 9(c), respectively. The “Langmuir C3 in which the residence time, (L/v), is ∼0.5 hour.
like” cases in Fig. 9(b) are not very good fits and the “non
Langmuir like” cases in Fig. 9(c) are very unlike the classic Table 2: The range of rate constants, κ , for the HW kinetic
Langmuir coverage in their shapes. However, we again note adsorption model (Eq. 14) and the corresponding Damköhler
that this dynamic adsorption isotherm is for the entire rock in numbers, N Da1 , used in the calculations in Figs. 11 – 13.
this case. It is possible that there are very small quantities of
Rate Time to Time to Notes
highly adsorbing minerals in this rock, which a multi-isotherm const. ⎛κL ⎞ 95% 99% (cases in
representation as in the G-F approach may describe quite well. N Da1 = ⎜ ⎟
κ ⎝ v ⎠ coverage coverage Figs. 11 – 13)
Indeed, the worst fitting case (Flood C3) was taken as shown
in Fig. 10 and then matched approximately with a multi- (1/s) (hour) (hour)
Langmuir isotherm of the form: 0.005 9.1 0.17 0.26 A. Effectively at eq.
4
i (c ) =
Γ ∑ β i Γi i
i =1
(19) 0.001 1.8 0.83 1.28 B. Very close to eq.
0.0005 0.91 1.66 2.56 C. Close to eq.
where the Γi i are the 4 Langmuir isotherms in this figure (α =
0.0001 0.18 8.32 12.8 D. Deviates from eq.
1, 0.1, 0.01 and 0.001) and the β i coefficients sum to 1.
Clearly, Fig. 10 shows reasonably good agreement between 1E-05 0.018 83.2 128 E. Far from eq.
the dynamic adsorption isotherm as found by the HW
procedure and a multi isotherm model as might be based on These results show for N Da1 ≈ 1 or greater (cases A, B and C),
the G-F approach as described above. the system is, as expected, close to equilibrium in that the long
tail behaviour of [SI] is essentially identical to the equilibrium
When Do Rate Effects Matter? adsorption case (Figs. 11 and 12). The deviation in the early
Adsorption rate, residence time and the Damköhler number: time, higher concentration effluents are observable, but small
It was noted in the text above, and shown in Fig. 1, that the (Fig. 13). Case D ( N Da1 ≈ 0.18 ) shows clear deviation from
different assumptions of the HW and G-F models on the equilibrium in the frontal and early time higher concentration
adsorption rate laws lead to different degrees of kinetic effects. effluents (Fig. 11 - 13), but is close to the equilibrium case in
The 2nd order rate law of Eq. 17 assumed by G-F leads to more the long tail at [SI] < 10 ppm. This low concentration
emphasis on these rate effects. However, this can be behaviour of case D may be surprising considering the 8 to 12
expressed more formally by considering the dimensionless hour coverage times (Table 2) as compared with the 0.5 hour
forms of the SI transport equation (Eq. 1) and the different rate fluid residence time in the core. However, the retardation due
6 SPE 95088

to the very steep isotherm (dΓ/dC >>1) effectively extends the evaluated with two different G-F composite isotherms. The
residence time in the core in the low concentration region isotherms are shown in Fig. 15 and compared to the original
(described by Eq. 5). Case E is at a low enough Damköhler HW isotherm. Isotherm 1 (dashed line) was constructed as a
number ( N Da1 ≈ 0.018 ) to ensure it is far from adsorption two-level approximate isotherm to the HW isotherm so that
equilibrium for the HW model to lead to strong kinetic effects equilibrium desorption conditions could be compared.
both in the early time, high concentration and in the later time, Isotherm 2 was constructed as a two-level composite isotherm
low concentration SI regions. using estimated low-pH parameters for an alumina-type
surface and a silica-type surface [6, 14].
Rate effects – G-F model: We now show that an effluent
profile can be matched with different sets of isotherm-kinetic Three simulations were performed: (1) Isotherm 1 under
desorption pairs. Flood F5 from previous work [3] was equilibrium desorption conditions, (2) Isotherm 2 under
conducted with DETPMP at neutral pH and so minimizes the equilibrium desorption conditions, and (3) Isotherm 2 with
number of processes occurring during desorption. The effluent kinetic desorption. The results from the simulations are
profile was only available for a small amount of total postflush reported in Fig. 16 as a log-log plot to magnify the region
volume. Still, three cases were used to demonstrate a fit using from 10 to 30 PV of effluent. Fig. 16 shows that Isotherm 1
did a reasonable job of matching the effluent, except in the
the G-F approach. Case 1 used an equilibrium constant (α) of
range of about 12 to 25 PV. However, this isotherm was not a
16,000 M-1 (Eq. 18) and equilibrium desorption. Case 2 used
perfect match to the HW isotherm, and so the discrepancy was
an equilibrium constant of 900 M-1 and a kinetic desorption
perhaps not unexpected. Isotherm 2 under equilibrium
rate constant of 20,000 (Eq. 17). Case 3 used an equilibrium
desorption conditions failed as expected. Notice that a large
constant of 300 M-1 and a kinetic desorption rate constant of
amount of scale inhibitor desorbed at about 50 ppm, which is
12,000. The plateau adsorption capacities for the 3 cases were
the approximate centre point of the lower level silica-type
located far above the SI concentration during column loading,
isotherm.
and so were also adjusted to achieve a good fit. For
comparison, Case 4 was also conducted using the Case 3
The third simulation used Isotherm 2 with kinetic desorption
isotherm and equilibrium desorption, though the results did
and gave an excellent match. The results are also reported in
not match the effluent profile. This last calculation
Fig. 17 in semi-log fashion for comparison to the HW
demonstrates the magnitude of impact that kinetic desorption
simulation in Fig. 8(a). Notice there was minimal impact on
might have on the early time SI concentrations.
the simulated effluent profile at high concentration and that
the major changes were in the low concentrations regime. This
The results of the calculations using these parameters are
behaviour is different from the HW approach only because of
shown in Fig. 14. Cases 1 - 3 provided acceptable fits of the
the “order” of the driving potential term, and so demonstrates
desorption profile but none could match the adsorption profile,
the utility that is possible with either kinetic model.
suggesting that perhaps chemical reactions might be occurring.
Indeed, the match of Case 1 using equilibrium desorption
The G-F and HW approaches tackle the same set of problems
suggests that adsorption should be 16,000 times faster than
from different viewpoints. We have shown that it is
desorption. If that were the case, then the effluent profile and
mathematically possible to construct a HW dynamic isotherm
the model should match very well. The discrepancy of the
by assembling pieces of separate isotherms using the G-F
modelling during adsorption provides some evidence of
approach. In addition, it is possible for either model to match
additional processes that warrant investigation. The
an effluent profile by assuming a static isotherm and adjusting
mathematical non-uniqueness of fit during desorption led to
the kinetics of desorption to agree with the effluent
the specific laboratory measurements of truly equilibrium
concentrations. It is also clear that a single set of G-F
adsorption isotherms and plateau adsorption capacities on
isotherms cannot capture all the complex effects embodied in
mineral separates. However, even this approach can contain
the HW dynamic isotherms. The fundamental question
uncertainties due to chemical reactions [20].
becomes whether the empirical dynamic isotherms obtained
by the HW approach in short-core flow tests are true
As mentioned earlier, the form of the desorption rate equation
equilibrium isotherms across their full range of determination.
for the G-F approach amplifies kinetic effects due to the
Future experiments may shed light on this question.
driving potential being a second-order (quadratic) term instead
of first-order. This can be seen more clearly in the low
Summary: In the calculations above, we have shown some of
concentration regime. For example, driving potentials of
the predicted rate effects of the HW and G-F models.
about “1” are unaffected by a second-order term, but driving
However, in practice, the correct form of the adsorption rate
potentials of “0.01” become “0.0001” with a second-order
equation (Eq. 14 or 17) may be established by performing
term. As such, very high potentials are made even faster, and
static adsorption rate experiments where the adsorption level is
very low potentials are made even slower with a second-order
monitored over time from some initial solution concentration,
term.
co, to some final value, cf. The integrated equations as shown
in Fig. 1 should then be compared with these experimental
The flow test for the low concentration regime comparison
results. This will help to establish the correct rate law as well
was the Flood C3 examined earlier, which was a low-pH
adsorption flood. The effluent SI profile from Flood C3 was as the actual values of the adsorption parameters (κ or kdes).
SPE 95088 7

This approach assumes, of course, that no additional chemistry mechanistic model, e.g. in adsorption/desorption treatments,
occurs, such as precipitation with calcium, extraction of no great preference has been made between Langmuir and
aluminum from clays to cause additional SI deposition, or Freundlich models [3, 9, 11]. Indeed, in the most general
other additional complex surface reactions. Chemical analysis in the HW approach, the adsorption isotherm, Γ(C)
reactions captured by an isotherm under one set of conditions (or pseudo-isotherm if the process is not really adsorption), is
in the laboratory may not transform properly to field derived as a table of numbers from a given set of experimental
conditions. Therefore, it will be important to identify and core flood data [3, 11]. As a result, the physical meaning of
isolate adsorption and chemical reaction effects. the resulting “isotherm” may be rather doubtful in some cases.
In the field, the same type of approach has been taken where
Discussion and Conclusions field SI return data has been used from a “typical” squeeze to
Comparison of HW and G-F models: We first give a brief derive the isotherm parameters (Langmuir or Freundlich)
summary of the differences between the HW and G-F scale which simply “fit the data”. This parameterised isotherm has
inhibitor squeeze modelling approaches [3, 6]. We have then been used to predict subsequent squeeze treatments in
considered these models in the light of their overall both the same well and in similar wells in that field [e.g. 8, 15-
mathematical structure and in the detailed manner in which 17]. Of course, the advantage of this approach is its
SI/rock adsorption is represented. We conclude that: generality, the fact that it does not take a strong mechanistic
view and that it can be applied quite straightforwardly by
(i) The mathematics are essentially identical, although using some well established software [18]. However, the
different units give slightly different forms of expression in disadvantage is that it is not truly predictive in that either
various papers; see Table 1. experimental data or field data must be matched to get Γ(C). A
partial solution to making this approach more predictive is to
(ii) However, there are three important differences in the use existing isotherms to predict behaviour in new reservoirs
details of the adsorption modelling between the two that are very similar.
approaches as follows:
- The form of the adsorption rate law is different in each G-F approach: An alternative approach taken by the
approach. A 1st order law is assumed in the HW model Halliburton group has been to start with a chemical view of
whereas a 2nd order rate law is assumed in the G-F approach. the inhibitor/mineral adsorption model for a given SI type
The latter emphasises the effect of rate, which depends on the (phosphonate)/mineral system [6, 14]. In this approach, the
ratio of adsorption rate to fluid velocity via the Damkohler adsorption isotherm has been specified as being of Langmuir
number (Appendix A); form (sometimes modified), since this has been found to
- The Langmuir form (sometimes modified Langmuir) of the adequately fit the experimental adsorption data for
equilibrium isotherm adsorption, Γeq (c) , is assumed by G-F phosphonate SI on a range of mineral separates such as
for the mineral separates (supported experimentally). The HW kaolinite, siderite, quartz, etc. [14]. Thus, if the mineralogical
leaves this open and either a Langmuir, Freundlich or table of composition of a given reservoir rock is known, a suitable
numbers (usually) may be used to model adsorption; combination of the various isotherms can be used (sometimes
- In the G-F approach, the entire rock is treated as a mineral in appropriate SI concentration ranges) to model the return
assemblage and multi-isotherm treatment may be applied by curves a priori. Thus, an advantage of this approach is that it
using a Langmuir form for each of the rock components. The may be used in predictive mode, if the appropriate SI/mineral
HW approach only considers a single adsorption isotherm data has been gathered and the rock mineralogical composition
although a clear connection between these approaches is is known. Some choices still have to be made with this
demonstrated for one of the dynamic cases. approach since the mineral/clay separates used to gather the
base case data may not have the same accessible surface area
Discussion: We have shown that the mathematics are in the consolidated mineral and hence some parameterisation
essentially “blind” to the precise mechanism by which we may be required here. A disadvantage of this approach is that a
view the retention mechanism. That is, the general form of the significant effort would be required to build a complete
transport equation is very similar irrespective of the actual database of all the various SI/mineral separates that covered
model describing the retention. Having said this, it is evident all specific brine compositions, formation temperatures, pH
in a given situation, the precise form of the retention model is values, etc. Analysis and fitting of field returns can, however,
very important to accurately “match” or “predict” the data, be provide the specific information needed to build such a
it a specific reservoir condition core flood or field return data. database for a field of operation. An advantage of this
Thus, it may be very important in detail whether we assume approach is that deviations of desorption behaviour from static
say a Langmuir or Freundlich adsorption isotherm or a kinetic isotherms in the database may indicate the opportunity to
precipitation model. discover new processes. Indeed, it was the deviation of
behaviour from isotherms on specific clays that drove the
HW approach: The early work of the HW group focused search for the relevant isotherm in very long tails. That search
primarily on the general mathematical form of the SI lead to the discovery of siderite, or iron-substituted carbonates,
retention/transport equations. The SI/rock interaction, as a major surface for providing the long tails. In this regard,
although it has been specified as adsorption/desorption or the HW and G-F approaches become very complementary.
precipitation, has not been strongly committed to a specific
8 SPE 95088

Closing remarks: Clearly, both of the approaches described 8. Jordan, M.M., Sorbie, K.S., Chen, P., Armitage, P., Hammond, P.
above are related and each has its advantages and and Taylor, K.: “The Design of Polymer and Phosphonate Scale
disadvantages. The main objective of this paper has been to Inhibitor Precipitation Treatments and the Importance of Precipitate
revisit and reconcile them in a general framework that clarifies Solubility in Extending Squeeze Lifetime,” SPE37275 presented at
the SPE International Symposium on Oilfield Chemistry, Houston,
all aspects of each in terms of both the mathematical structure TX, 18-21 February 1997.
of the equations and the details of the SI/rock retention model.
There are differences between these two models in terms of 9. Sorbie, K.S., Wat, R.M.S., Todd, A.C. and McClosky, T.:
their static adsorption models and the non-equilibrium rate “Derivation of Scale Inhibitor Adsorption Isotherms for Sandstone
laws. However, all of the assumptions used in each of the Reservoirs,” Royal Society of Chemistry Publication - Chemicals in
models for a given SI/rock/brine system are open to the Oil Industry: Developments and Applications, edited by P. H.
experimental verification. In the future, we will be devising Ogden, 1991.
various laboratory experiments to test the models used in each
of these approaches. 10. Sorbie, K.S., Yuan, M.D., Chen, P., Todd, A.C. and Wat, R.M.S.:
“The Effect of pH on the Adsorption and Transport of Phosphonate
Scale Inhibitor Through Porous Media”, SPE25615, presented at the
Acknowledgements SPE International Symposium on Oilfield Chemistry, Anaheim, CA,
The authors would like to thank the sponsors of the Flow 20-22 February 1993.
Assurance and Scale Team (FAST) at Heriot-Watt University:
FAST I sponsors - Baker Petrolite, BioLab, BP, 11. Yuan, M.D., Sorbie, K.S., Jiang, P., Chen, P., Jordan, M.M.,
ChampionServo, ChevronTexaco, Clariant Oil Services, Todd, A.C., Hourston, K.E. and Ramstad, K.: “Phosphonate Scale
ConocoPhillips, Exxon/Mobil, Halliburton, Kerr-McGee, Inhibitor Adsorption on Outcrop and Reservoir Rock Substrates - The
Marathon, MI Production Chemicals, Norsk Hydro, Nalco, ‘Static’ and ‘Dynamic’ Adsorption Isotherms", in Recent Advances in
Petrobras, Rhodia, Saudi Aramco, Schlumberger, Shell, Oilfield Chemistry, Edited by P. H. Ogden, Royal Society of
Chemistry, Special Publication No. 159, 1994.
Statoil, Total and Yukos. They also thank Halliburton Energy
Services, Inc. for permission to publish this work. 12. H. Zhang, H., Mackay E.J., Chen P. and Sorbie K.S.: “Non-
Equilibrium Adsorption and Precipitation of Scale Inhibitors:
Corefloods and Mathematical Modelling”, SPE64755, Proceedings of
References SPE International Oil and Gas Conference and Exhibition, Beijing,
1. Meyers, K.O., Skillman, H.L. and Herring, G.D.: “Control of China, 7–10 November 2000.
Formation Damage at Prudhoe Bay, Alaska by Inhibitor Squeeze
Treatment, SPE12472, J. Pet. Tech., pp. 1019-1034, June 1985. 13. Sorbie, K.S., Jiang, P., Yuan, M.D., Chen, P., Jordan, M.M. and
Todd, A.C.: “The Effect of pH, Calcium and Temperature on the
2. Hong, S.A. and Shuler P.J.: “A Mathematical Model for the Scale Adsorption of Phosphonate Inhibitor onto Consolidated and Crushed
Inhibitor Squeeze Process”, SPE16263, SPE (Production Sandstone,” SPE26605 presented at the SPE Annual Conference and
Engineering), pp. 957-607, November. 1988. Technical Exhibition, Houston, TX, 3-6 October 1993.

3. Sorbie, K.S., Wat, R.M.S. and Todd, A.C.: “Interpretation and 14. Funkhouser, G.P. and Gdanski, R.D.: “Improved Adsorption-
Theoretical Modelling of Scale-Inhibitor/Tracer Corefloods,” SPE Isotherm Modeling for Phosphonate Scale Inhibitors”, Presented at
Production Engineering, pp. 307-312, August 1992. the NIF International Oilfield Chemistry Symposium, Geilo, Norway,
1-4 April 2001.
4. Sorbie, K.S., Yuan, M.D. and Todd, A.C.: “The Modelling and
Design of Scale Inhibitor Squeeze Treatments in Complex 15. Yuan, M.D., Sorbie, K.S., Todd, A.C., Atkinson, L.M., Riley, H.
Reservoirs”, SPE21024, presented at the SPE International and Gurden, S.: “The Modelling of Adsorption and Precipitation
Symposium on Oilfield Chemistry, Anaheim, CA, 20-22 February Scale Inhibitor Squeeze Treatments in North Sea Fields,” SPE25163
1991. presented at the SPE International Symposium on Oilfield Chemistry,
New Orleans, LA, 2-5 March 1993.

5. Kan, A., Cao, X., Yan, P.B., Oddo, J.E. and Tomson, M.B.: 16. Børeng, R., Sorbie, K.S. and Yuan, M.D.: "The Underlying
“The Transport of Chemical Scale Inhibitors and its Importance to Theory and Modelling of Scale Inhibitor Squeezes in Three Offshore
the Squeeze Procedure,” paper No. 33 presented at the NACE Annual Wells on the Norwegian Continental Shelf", Proceedings of the Fifth
Conference and Corrosion Show, Nashville, Tennessee, 27 April - 1 International Oilfield Chemicals Symposium, Geilo, Norway, 20-23
May 1992. March 1994.

6. Gdanski, R.D. and Funkhouser, G.P.: “Successful Model of the 17. Sorbie, K.S., Yuan, M.D., Jordan, M.M. and Hourston, K.E.:
Kinetic Release of a Phosphonate Scale Inhibitor”, Presented at the “Application of a Scale Inhibitor Squeeze Model to Improve Field
NIF International Oilfield Chemistry Symposium, Geilo, Norway, 1- Squeeze Treatment Design”, SPE28885, Proceedings of the SPE
4 April 2001. European Petroleum Conference (Europec 94), London, UK, 25-27
October 1994.
7. Malandrino, A., Yuan, M.D., Sorbie, K.S. and Jordan, M.M.:
“Mechanistic Study and Modelling of Precipitation Scale Inhibitor 18. Zhang, H. and Sorbie, K.S.: The SQUEEZE V User Manual,
Squeeze Processes”, SPE29001, Proceedings of the SPE International Institute of Petroleum Engineering, Heriot-Watt University,
Symposium on Oilfield Chemistry, San Antonio, TX, 14-17 Edinburgh, 1997.
February 1995.
SPE 95088 9

19. Gdanski, R.D. and Funkhouser, G.P.: “Fully Contained Table A1: The formulae for the various dimensionless
Laboratory Squeeze Treatments”, Presented at the NIF International numbers in Eqs. A1 – A3 and their physical meaning.
Oilfield Chemistry Symposium, Geilo, Norway, 17 – 20 March 2002. Dimension Expression Physical meaning
-less
20. Gdanski, R.D. and Funkhouser, G.P.: “Mineralogy Driven Scale
Inhibitor Squeeze Designs”, SPE94510 to be presented at the number
European Formation Damage Conference, Scheveningen, The ⎛ vL ⎞ Ratio of convection to
Netherlands, 25 – 27 May 2005. Peclet No. N Pe = ⎜ ⎟ dispersion
⎝D⎠ (mixing/spreading);
High N Pe – convection
Appendix A: The Dimensionless SI Transport Equations dominated flow; Low
In order to appreciate the balance between the various physical N Pe – dispersion
effects that occur when SI is transported through a porous
dominated flow.
medium, we must cast the relevant equations in dimensionless
form. The effects in question are – convection or rate of ⎛ ρΓ ⎞ Ratio of adsorptive
transport (v), dispersion (D), adsorption level ( Γ max ) and
Langmuir N La = ⎜ max ⎟ capacity of the rock to a
No.
⎝ φ c0 ⎠ PV of mobile SI at input
adsorption rate (κ and k des in Eqs. 14 and 15). Defining concentration, c0 . The
dimensionless length (X = x/L), time (T = tv/L) and adsorption larger is N La , the more
i = Γ (c) / Γ ),
level ( Γ max dominant adsorption is
Eq. 1 becomes: and the more frontal
⎛ ∂C ⎞ 1 ⎛ ∂ 2C ⎞ ⎛ ∂C ⎞ i⎞
⎛ ∂Γ retardation is observed.
(A1)
⎜ ⎟= ⎜ 2 ⎟−⎜ ⎟ − N La ⎜ ⎟
⎛κL ⎞ Ratio of adsorption rate
⎝ ∂T ⎠ N Pe ⎝ ∂X ⎠ ⎝ ∂X ⎠ ⎝ ∂T ⎠ Damköhler N Da1 = ⎜ ⎟ to fluid convection rate.
No. ⎝ v ⎠ The higher is N Da , the
Eq. 14 becomes: for Eq. 14 - A2 or
i⎞ closer the system is to
⎛ ∂Γ
⎜ ⎟ = N Da1 Γ (
i eq (C ) − Γ
i (T ) ) (A2)
⎛ k .L.Γ 2max ⎞
equilibrium adsorption,
⎝ ∂T ⎠ i =Γ
Γ i eq (c) .
N Da 2 = ⎜ des ⎟ i.e.

and Eq. 15 becomes: ⎝ v ⎠

i⎞ for Eq. 17 – A3
⎛ ∂Γ
⎜ ⎟ = N Da 2 ( i eq (C ) − Γ
Γ )
i (T ) 2 .Γ
i (T ) (A3)
⎝ ∂T ⎠

C mobile vs. time

The three dimensionless numbers, N Pe , N La and N Da are the
Peclet, Langmuir, and Damköhler numbers, respectively. The 0.14 Rate law
S. rate – Eq.
equn. 14 =(κ0.03)
(kappa = 0.03)
Damköhler number has two variants, N Da1 and N Da 2 ,
0.12
Mobile phase conc. C

corresponding to the two non-equilibrium adsorption

0.1
expressions [3, 14]. These quantities and their physical Series1
0.08
meaning are given in Table A1. G-F rate equn. (k =2)
Series2
0.06 Rate law – Eq. 17
15 (k = 2.0) Series3

0.04

0.02

0
0 200 400 600 800 1000
Time,tTÆ
Time, -->

Fig. 1: The kinetic adsorption (or desorption) rate laws of the

HW (Eq. 14, [3]) and G-F (Eq. 17, [6]) models integrated
(numerically) for the same solution conditions for a static case.
All quantities are dimensionless.
10 SPE 95088

1
1.2

0.9
1.0
0.8

Fractional Adsorption
α = 1.0
0.7 0.8
Fraction coverage

0.6
α = 0.1 0.6
0.5

0.4
α = 0.01
-1
0.4 Keq = 50,000 M
0.3 C50 = 3.5 µM
β = 3.5
0.2 α = 0.001 0.2 capacity = 3.0 µmol/m
2

0.1

0 0.0
0.1 1 10 100 1000 10000 0.000001 0.00001 0.0001 0.001 0.01
SI concentration (ppm) Supernatant Concentration, M
Fig. 5: Experimental data for the adsorption of DETPMP on
siderite; from Funkhouser and Gdanski [14].
Fig. 2: Fractional coverage ( Γi ) vs. SI conc., c (ppm) for a
Langmuir adsorption isotherm with α values 1.0, 0.1, 0.01 and
0.001. Note that Γi = 0.5 at c = 1/α.
8000
1.2 7000

Adsorption (mg/L)
6000 Flood C3
1.0 Flood C4
5000
Fractional Adsorption

Flood C7
0.8 4000 Flood C10
Flood C11
3000
0.6 Flood C13
2000

0.4 Keq = 15,900 M

-1 1000
C50 = 62.7 µM
0
β=0
0.2 capacity = 0.15 µmol/m
2 0 1000 2000 3000 4000 5000
SI concentration (ppm)
0.0
0.000001 0.00001 0.0001 0.001 0.01 0.1
Supernatant Concentration, M Fig. 6: Experimental dynamic isotherms Γ vs. c (ppm) for
Fig. 3: Experimental data for DETPMP on kaolinite at pH 7 phosphonate floods (C3, C4, C7, C10, C11 and C13) in
fits by the Langmuir curve; from Funkhouser and Gdanski reservoir cores; note – units are in mg SI/L of rock grain (ΓE
[14]. in Table 1) where 1000 mg/L = (1−φ)/ρ mg/g which for this
rock, 1000 mg/L = 0.303 mg/g or 7000 mg/L = 2.12 mg/g.
1
0.9 8000

0.8
SI Adsorption (mg/g)

7000 Flood C3
0.7 6000
Flood C4
Adsorption (mg/L)

Flood C7
0.6
5000 Flood C10
0.5 bα= =
1 1.0 Flood C11
4000
0.4 α = 0.1
b=0.1
Flood C13
α 0.01
b= = 0.01 3000
0.3
α = 0.001
b=0.001
0.2 2000

0.1 1000
0 0
0 1000 2000 3000 0.1 1 10 100 1000
SI concentration (ppm) SI concentration (ppm)

Fig. 7(a)
Fig. 4: Isotherms Γ (mg/g) vs. SI conc. c (ppm) for a
Langmuir adsorption isotherm with Γmax= 1 (mg/g) for α
values 1.0, 0.1, 0.01 and 0.001 in Fig. 2.
SPE 95088 11

8000 1

0.9
7000
0.8
6000 Flood C3
Adsorption (mg/L)

Flood C4 0.7 bα==11.0

Fraction coverage
5000
Flood C7 0.6 α = 0.1
b=0.1
4000 Flood C10 α =0.01
b= 0.01
0.5
Flood C11 α =0.001
b=0.001
3000
Flood C13 0.4 Flood C3
2000 Flood C4
0.3
Flood C7
1000 0.2 Flood C10
Flood C11
0 0.1
Flood C13
0 50 100 150 200 250 300
0
SI concentration (ppm) 0.1 1 10 100 1000 10000
SI concentration (ppm)
Fig. 7(b)
Fig. 9(a): All dynamic adsorption isotherms
Fig. 7: The same experimental dynamic isotherms Γ vs. c
(ppm) for phosphonate floods (C3, C4, C7, C10, C11 and 1
C13) as in Fig. 6. with 7(a) log concentration scale and 7(b) 0.9
linear scale, but in lower concentration range of 0 – 300 ppm. 0.8

0.7

Fraction coverage
0.6
αb == 1.0
1
0.5 αb=0.1
= 0.1
αb== 0.01
0.01
0.4
αb=0.001
=0.001
0.3 Flood C4
Flood C7
0.2
Flood C11
0.1

0
0.1 1 10 100 1000 10000
SI concentration (ppm)

Fig. 8(a): Flood C3 Fig. 9(b): Langmuir like

0.9

0.8

0.7
Fraction coverage

0.6
bα==11.0
0.5 α = 0.1
b=0.1
α =0.01
b= 0.01
0.4
α =0.001
b=0.001
0.3 Flood C3
Flood C10
0.2
Flood C13
0.1
Fig. 8(b) Flood C4 0
0.1 1 10 100 1000 10000
Fig. 8: Simulation of core floods based on the adsorption SI concentration (ppm)
isotherms shown in Fig. 6 (and 7) showing both early time and Fig. 9(c): Non Langmuir like
long tail predictions; (a) Flood C3 and (b) Flood C4; points =
experiment, solid line = simulation. i (c ) ) for the
Fig. 9: Comparison of the fractional coverage ( Γ
dynamic adsorption isotherms in Fig. 6 (and 7) with the ideal
Langmuir fractional coverage in Fig. 2; (a) all cases, (b)
cases where the coverage appears quite “Langmuir like”, C4,
C7 and C11, and (c) cases where the coverage is “non
Langmuir like”, C3, C10, C13.
12 SPE 95088

1
A
0.9
20000 D
0.8 C
E
0.7 Flood C3
Fraction coverage

15000
0.6 Composite
isotherm fit

[SI] (ppm)
0.5
bα= = 1.0
1.0 10000 B - rate const. = 0.005 (1/s)
0.4
C - rate const. = 0.001 (1/s)
bα= = 0.1
0.1
0.3 A to E in D - rate const. = 0.0001 (1/s)
this order E - rate const. = 0.00001 (1/s)
bα==0.01 5000
0.2 0.01 A - at equilibrium

0.1 α ==0.001
b 0.001
0
0
0 5 10 15 20
0.1 1 10 100 1000 10000 PV injected (PV)
SI concentration (ppm)

Fig. 13: Running the equilibrium adsorption model (A) for

Fig. 10: Composite isotherm fit (solid line) to the dynamic Flood C3, but allowing kinetic effects: κ = 0.005 (B), 0.001
isotherm for core flood C3 (-*-) using a combination of the (C), 0.0001 (D) and 0.00001 s-1 (E) in Eq. 14. Entire [SI] range
four Langmuir forms shown. over early fluid PV injected (0 Æ 20 PV).

100000
3000
A - At equilibrium Observed
B - rate const. = 0.005 (1/s) Case 1
10000
C - rate const. = 0.001 (1/s)
2500 Case 2
Case 3
D- rate const. = 0.0001 (1/s) Case 4
1000
Inhibitor (ppm)

E - rate const. = 0.00001 (1/s) 2000

[SI] (ppm)

D
100
1500
E Case 4
10
1000
E
1
A, B and C almost identical 500
over this SI conc. range
0.1
0
0 100 200 300 400 500 600 700 800
PV injected (PV) 0 50 100 150 200
Effluent Volume (mL)
Fig. 11: Running the equilibrium adsorption model (A) for
Flood C3, but allowing kinetic effects: κ = 0.005 (B), 0.001 Fig. 14: Flood F5 [3] desorption matching for G-F model
(C), 0.0001 (D) and 0.00001 s-1 (E) in Eq. 14. Entire [SI] range with 4 conditions of isotherm and kinetics. Case 1: α = 16,000
over entire fluid PV injected. M-1, equilibrium desorption; Case 2: α = 900 M-1 and k des =
20,000; Case 3: α = 300 M-1, k des = 12,000; Case 4: Case 3
100
A - At equilibrium isotherm and equilibrium desorption.
B - rate const. = 0.005 (1/s)

C - rate const. = 0.001 (1/s) 1.0

D- rate const. = 0.0001 (1/s) HW Isotherm

E - rate const. = 0.00001 (1/s)

D 0.8
[SI] (ppm)

E
Fractional Coverage

10

0.6

A, B and C almost
identical over this SI 0.4
Isotherm 2
conc. range

0.2
1 Isotherm 1
Loading
0 100 200 300 400 500 600 700 800 Conc.
PV injected (PV)
0.0
0.1 1. 10. 100. 1,000. 10,000. 100,000.
Fig. 12: Running the equilibrium adsorption model (A) for Scale Inhibitor (ppm)

Flood C3, but allowing kinetic effects: κ = 0.005 (B), 0.001 Fig. 15: Flood C3 potential isotherms: original HW isotherm
(C), 0.0001 (D) and 0.00001 s-1 (E) in Eq. 14. Lower range, and two candidate G-F composite isotherms.
[SI] < 100 ppm, over entire fluid PV injected.
SPE 95088 13

100,000 100,000

10,000 10,000
Inhibitor (ppm)

Inhibitor (ppm)
1,000 1,000
Isotherm 1
Equilibrium

100 100
Isotherm 2
Equilibrium
10 10
Isotherm 2
kdes = 2E3

1 1
1 10 100 1,000 0 100 200 300 400 500 600 700 800
Effluent Volume (PV) Effluent Volume (PV)
Fig. 16: Flood C3 desorption matching for G-F model with 3 Fig. 17: Flood C3 desorption matching with G-F Isotherm 2
conditions of isotherm (as in Fig. 15) and kinetics, as follows: (Fig. 15) and kinetic desorption.
Isotherm 1 under equilibrium desorption conditions; Isotherm
2 under equilibrium desorption conditions; Isotherm 2 with
kinetic desorption.