Journal of Petroleum Science and Engineering 190 (2020) 107048

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering

journal homepage: http://www.elsevier.com/locate/petrol

Tracer eluting proppants for hydraulic fracture characterization

Bochao Zhao , Krishna Panthi , Kishore K. Mohanty *
Hildebrand Department of Petroleum & Geosystems Engineering, The University of Texas at Austin, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Existing fracture characterization methods (e.g., microseismic) detect fractures created during hydraulic frac­
Tracer turing, but cannot determine whether the fractures are propped or not. The goal of this work is to develop a
Proppant tracer-eluting proppant technique that can indicate the extent of propped fractures. A polymer coating is
Fracture characterization
developed for common proppants which can release a tracer after being placed in a fracture. A small slug of
Polymer coating
Hydraulic fracturing
coated proppants can be injected at the end or middle of the regular proppant injection during hydraulic frac­
turing. The released tracer can be collected during flow back and be used to better understand proppant location
in fracture systems. Laboratory-scale batch tracer release experiments show that the tracer release can be trig­
gered by an increase in temperature, salinity or pH. A proof of concept experiment was conducted in a model
single fracture system. The tracer was successfully detected in the flow-back fluid and the tracer concentration
was accurately estimated by a numerical tracer release model. The location of the tracer eluting proppants can be
estimated from the peak tracer time in the flow-back water. This technique can be used for a wide range of
proppants and tracers. Different tracers can be used in different stages.

1. Introduction treatment well (Roussel and Agrawal, 2017; Seth et al., 2018). This
method is useful for understanding the communication between wells
Unconventional reservoirs such as shales need to be hydraulically and stages, but it cannot distinguish between propped and unpropped
fractured to produce oil and gas at an economic rate. Long propped fractures, similar to the microseismic method.
hydraulic fractures connected to pre-existing natural fractures are Chemical tracers have also been used for unconventional reservoirs.
needed to maximize productivity for ultra-low permeability shales Traditionally chemical tracers have seen widespread use in secondary
(Weng et al., 2011). Fracture geometry (such as position and length) and and tertiary recovery for conventional reservoirs for determination of
conductivity affect well productivity (Warpinski et al., 2008). It is also pore volume, swept volume and residual oil saturation (Robinson et al.,
important to characterize the created fracture system for hydraulic 1988; Shook et al., 2009; Tang and Harker, 1991). Recently, chemical
fracturing process optimization. tracers have also been applied to unconventional shale reservoirs to
The following methods exist for hydraulic fracture characterization: assess fracture connectivity and evaluate stimulation effectiveness
seismic, tracers and logging. Microseismic monitoring is commonly (Salman et al., 2014; Gardien et al., 1996). Chemical tracers can be
conducted during hydraulic fracturing (Fisher and Warpinski, 2012; either water soluble (Asadi et al., 2002; King and Leonard, 2011) or oil
Maxwell et al., 2002). The location of microseismic event can be used as soluble (Catlett et al., 2013; Goswick and LaRue, 2014). The tracer is
a proxy for slip failure of fractures or faults that usually occurs during introduced into the reservoir with the fracturing fluid or incorporated in
creation and propagation of a fracture. This method has a long detection the perforation process (or placed outside of liner downhole). After
distance and can create a time dependent map of the fracture system hydraulic stimulation, the tracer is collected along with the flow-back
(Zoback et al., 2012). However, this method does not indicate if the fluid and its concentration is analyzed. The shape and value of the
created fractures remain open (Eisner et al., 2011). Microseismic ac­ flow-back curve can be used to determine fracture properties. The lim­
tivity could be affected by the in-situ stress and pore pressure; the itation of this method is that tracer data is subject to significant
fractures created by tensile opening are not detected. Pressure inter­ dispersion in the fracture system and quantitative interpretation of
ference can also be used to understand hydraulic fracture geometry by tracer data is challenging (Li et al., 2016). One limitation of chemical
observing the pressure changes in adjacent wells during fracturing in a tracer is that the tracer can be transported into un-propped fractures and

* Corresponding author.
E-mail address: [email protected] (K.K. Mohanty).

https://doi.org/10.1016/j.petrol.2020.107048
Received 3 January 2020; Received in revised form 5 February 2020; Accepted 7 February 2020
Available online 13 February 2020
0920-4105/© 2020 Elsevier B.V. All rights reserved.
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

fractures that are open but not connected to the wellbore. This has 2. Methodology
shown to cause difficulties in the interpretation of tracer test results as
multiple peaks with varying peak location can arise due to unpropped 2.1. Materials
fracture closure (Kumar and Sharma, 2018).
The previous methods fail to consider proppant transport during 40/70 mesh Ottawa sand was used as the proppant. Methacrylic acid
fracturing. Fractures propped with proppant remain open when pressure – ethyl acrylate copolymer was used as the polymer. Rhodamine 6G,
is reduced (after fracturing) and contribute to production. Some of the ethanol (>99.5%) and sodium chloride were used as received. Deionized
fractures that are opened during stimulation, but unpropped, close. water (with a resistivity greater than 18.2 MΩ-cm) was used to prepare
Some unpropped fractures remain open due to shear displacement, but aqueous solutions. The pH of the various solutions was measured using
have low conductivity. Previous methods give information about frac­ pH meter with the precision of �0.01. The pH electrode was calibrated
ture system during stimulation, but not the final fracture system when with standard pH buffer solutions of pH 4, 7, and 10. Buffer used to
pressure is dropped and production begins. For example, slickwater with control pH includes: pH 8.0 buffer (Potassium phosphate monobasic,
sand is used commonly in hydraulic fracturing of shales, because it sodium hydroxide), pH 9.0 buffer (Boric acid, potassium chloride, so­
creates long, skinny fractures (and is relatively inexpensive). One of the dium Hydroxide Buffer, pH 10.0 buffer (Potassium carbonate, potassium
main concerns during the slickwater fracturing process is the proppant hydroxide, potassium borate), pH 11.0 buffer (Boric acid, sodium hy­
transport in fractures and fracture intersections. Due to higher density, droxide, potassium chloride), pH 12.0 buffer (Sodium hydroxide, so­
sand settles and forms a sand bed; newly injected sand moves at the top dium phosphate).
of the sand bed and is deposited downstream (Tong and Mohanty,
2016). It is important to know the placement of proppant in addition to 2.2. Tracer-eluting proppant fabrication
the geometry of the created fracture system. Neither microseismicity nor
tracer can provide this information. 5 wt% polymer is first dissolved in acetone. The chemical tracer of
One potential method to directly detect proppants in the fracture is choice is then dissolved in the polymeric solution. Rhodamine 6G was
through logging. Chemicals can be embedded into proppants and selected as the tracer for its easy detection with an UV–vis spectrometer,
detected by logging equipment, thus giving information about prop­ but other tracers can be used as long as the tracer can be dispersed in the
pants distribution in fractures (McDaniel et al., 2007). These chemicals polymeric solution and is soluble in water. The proppant of choice is
can be radioactive (King and Leonard, 2011) or semi-radioactive (Bhatia then put into the solution and mixed thoroughly. 40/70 mesh Ottawa
and Pande, 2016; McDaniel et al., 2009). However, this technique is sand was used in this work, but other proppants can be used. Then the
limited by the radius of investigation of the logging tool, which is usually polymeric solution is drained out, leaving the proppants and the liquid
a few feet. Fractures can be hundreds of feet long, far out of the detection trapped around the proppants. The proppants are then placed in a well-
range of normal logging operation (Cipolla and Wright, 2002). In ventilated hood for the acetone to evaporate. When the acetone evap­
addition to radioactive proppants, electrically conductive proppants orates, the polymer forms a layer of coating around the sand surface,
have also been proposed to directly map proppant distribution using while trapping some tracer inside the coating. In this way, tracer is
electromagnetic borehole measurement for both open-hole and encapsulated in the polymer coating around the proppant. The tracer
steel-cased wellbores (Basu and Sharma, 2014; Zhang et al., 2018). This loading can be tuned by changing the amount of tracer dissolved in the
method has the advantage that the proppant is non-radioactive and the polymer solution. The entire batch process can be finished in 1 h and no
detection radius of this method could be on the order of 100 ft. However, chemical reaction or heating is required.
for through-casing detection, this method requires downhole installa­
tion of insulating gap and the signal is dominated by the fracture that is 2.3. Tracer elution tests
in directly in contact with the electrode on the casing.
The drawback of directly injecting chemical tracers into a well is that Ideally, the tracer should remain in the coating on the proppant
the tracer cannot accurately reflect proppant distribution in the system. during proppant injection in fractures until the proppants settle
If a tracer can be incorporated on the proppant and be released after completely. After proppant placement, the tracer should be released. If
proppant placement, then information about proppant location can be the tracer is released during injection period, then it cannot give effec­
estimated. This idea imposes two requirements on the tracer system. tive information about the final location of proppants. On the other
First, the proppant should not elute tracer at before and during proppant hand, if the tracer is released too slow after proppant placement, a low
injection. Minimum amount of tracer should be released at these two tracer concentration in the flow-back liquid may pose difficulty in the
stages. Second, the tracer should be released in sufficient amount after detection process. Therefore, the ability of the proppant to retain and
proppant placement under the reservoir condition or certain triggering release the tracer is critical in its applicability. To study this, we per­
conditions. These two requirements are analogous to the controlled formed batch tracer release tests.
release technology used in medicine (Bae et al., 2005; Siepmann et al., In these tests, coated proppants were mixed with a certain volume of
2008; Muhammad et al., 2011). Delayed release of oxidative gel brea­ a liquid (such as brine at a certain pH) at a constant temperature. A small
kers has been used in hydraulic fracturing (Watson et al., 2010) and sample of the liquid was taken periodically and its UV–vis absorbance
controlled acid release has been applied for acid treatments (Johnson was measured. Rhodamine 6G has a significant absorption peak around
et al., 2016). The controlled release technology is applied here for the 530 nm. Based on the absorption value, rhodamine concentration can be
first time for diagnosis of hydraulic fracturing. calculated from a standard curve. The amount of tracer released was
In the following, we describe a novel tracer-eluting proppant tech­ calculated as a function of time. Tracer-embedded proppants were
nology for estimation of proppant distribution in a hydraulic fracture immersed in DI water at 4 different temperatures: 25 � C, 60 � C, 90 � C
system. In this technology, a chemical tracer is encapsulated inside a pH and 125 � C and the elution was studied. The pH of water was changed to
sensitive polymer coating on proppants. These proppants can be injected 8, 9, 10 and 11 to study the effect of pH. The salinity of the water was
along with normal proppants. When proppants settle inside the fracture, varied to 50,000 ppm, 100,000 ppm and 150,000 ppm to study the effect
tracer can be released due to the reservoir temperature, salinity or pH of salinity.
conditions and be collected in the flow-back fluid. In this paper, we
evaluate the tracer eluting properties of these novel proppants and es­ 2.4. Fracture cell tracer elution tests
timate their position in the fracture from the flow-back water analysis.
The batch test can be used to identify the release condition for tracer
eluting proppants. After the batch test, dynamic tracer release tests were

2
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

conducted in a model fracture system. The fracture system was con­ function of time in Fig. 2 for each temperature. The total tracer loading
structed using a 2-inch diameter and 1 ft long Texas Cream limestone was calculated by placing the proppants into a pH 12 buffer (dissolves
core. The core was cut in half longitudinally along the central axis and the polymer and releases all of the tracer) and measuring the tracer
sand was placed between the cut surfaces to serve as proppants. Then the concentration. The average Rhodamine loading on proppant was 0.5 mg
core was put back together, wrapped in a heat-shrink film and put into a of Rhodamine per gram of proppant. The water to proppant ratio in all
Hassler coreholder. The Texas Cream limestone core acts as the matrix batch release experiments was between 400 and 500 gm of water per gm
and the space between the half core acts as the fracture propped by of proppant. The maximum Rhodamine concentration in solution was
proppants. The fracture volume (FV) was obtained and validated using 1.5 ppm. The detection limit of rhodamine in solution was 0.1 ppm,
two methods. In the first method, FV was obtained by multiplying the which ensures an accurate detection of the tracer concentration.
average fracture width with the fracture area and then subtracting the At 25 � C and 60 � C, the tracer was not detected in a 24-h period. In
proppant volume. In the second method, a salinity tracer test was con­ fact, no tracer was detected even after 1 week. This shows that the
ducted in the fractured core. In this tracer test, the fracture was first tracer-eluting system is very stable at a low salinity and low tempera­
saturated with 1% NaCl brine, then 6% NaCl brine was injected and the ture, making the transportation and storage of these proppants easy. As
effluent salinity was analyzed. The fracture volume is estimated to be the the temperature increased, the tracer started to get released. At 90 � C,
volume corresponding to 0.5 normalized concentration (after consid­ approximately 20% of the tracer loaded was released in a 24-h period. At
ering the dead volume). In all experiments, these two methods resulted 125 � C, almost all of the tracer was released within 10 h. A faster release
in consistent fracture volume estimations. The fracture volumes for the at higher temperature is expected because of increased diffusion of the
experiments 1, 2 and 3 were 12, 13 and 13.5 ml, respectively. tracer in the coating and mass transfer into the surrounding water.
For the dynamic tracer release test, a certain amount of tracer-eluting
sand was placed at specific locations on the fracture while ordinary sand 3.3. Tracer elution by pH change
was used to cover the rest of the fracture. The core and the fracture space
were pre-saturated with 1% NaCl brine. Next, a slug of pH 12 buffer The proppants were first immersed in DI water. Then, a pH buffer
(with the volume of the fracture space) was injected to trigger the was added to the solution. The pH of the solution was maintained at a
release of tracer on tracer-eluting proppants. After injection, the buffer certain value adding more buffer, if necessary. The tracer release is
flow was stopped for 12 h to mimic the shut-in time in the field and to shown in Fig. 3. For the experiment at 90 � C, the buffer was added at
allow sufficient time for tracer release. Finally, 1% NaCl brine was around t ¼ 21 h. For the experiments at 25 � C, 60 � C and 125 � C, the
injected from the opposite direction to mimic the flow-back period and buffer was added at t ¼ 0. Note that even though the buffer was added,
the effluent from the fractured core was analyzed for the tracer. Three the overall salinity of the solution was still very low. At 25 � C, 60 � C and
experiments with different tracer-eluting sand loading and location were 90 � C, increasing pH above 9 caused the tracer to be released
conducted, as shown in Table 1. completely. The time needed for complete release was around 10 h, 5 h
and 1 h under 25 � C, 60 � C and 90 � C, respectively again indicating an
3. Experimental results increase in the tracer release with temperature. A pH ¼ 8 was insuffi­
cient for a rapid release of the tracer. At pH ¼ 8, only around 10–20% of
3.1. Polymer-coated proppants the total tracer was released in 2 weeks at 25 � C, 60 � C and 90 � C. At 125
�
C, increasing the pH led to a rapid release of the tracer, especially for
Optical micrograph of the proppants before and after coating with pH ¼ 9 and 10. However, as indicated in the previous section, even at
the polymer containing Rhodamine 6G (tracer) is shown in Fig. 1. neutral pH, the tracer is completely released in 10 h. Additionally, it
Rhodamine 6G has a purple color in the visible wavelength. Coated sand seems that the level of pH has a less significant effect on the rate of tracer
grains gained a purple color, indicating a coating of tracer containing release compared with temperature (which is the dominant factor). A pH
polymer on the proppant surface. To estimate the average coating of 9 is often sufficient for a rapid and complete release of the tracer.
thickness, first, the dry coated-proppants were weighed. Then, they were Further increase in pH to 10 or 11 leads to an increase in the release rate,
put into an alkaline solution (pH > 12) to dissolve the polymer coating. but the increase is not very significant.
The proppants were then dried and weighed again. Based on the weight
difference and polymer density, average coating thickness was calcu­
3.4. Tracer elution by salinity change
lated to be 3.8 μm, which is only 1–2% of the sand grain radius. Such a
thin coating ensures minimal changes in proppant density due to
All previous experiments were conducted at a low salinity (similar to
coating. This is crucial as it ensures that tracer-coated proppant is a good
fresh fracturing water). However, many of the shale reservoirs have
representation of the proppants injected. The proppants dispersed in
highly saline formation water. It is necessary to study the effect of
water (and did not stick together), indicating that the proppants were
salinity on tracer release for determining whether tracer-eluting prop­
water-wet after coating.
pants technology can be applied to these reservoirs. In this experiment,
proppants were placed in NaCl solution of different salinity: 50,000
3.2. Tracer elution in DI water ppm, 100,000 ppm and 150,000 ppm instead of DI water at different
temperatures. The results are plotted in Fig. 4. At the room temperature,
Tracer-eluting proppants are placed in a bath of DI water at a con­ increasing the salinity led to 10–20% tracer release over a 150-h period.
stant temperature and the tracer concentration in the water was moni­ However, the release stopped and did not increase further with time. At
tored. The fraction of the embedded tracer released is shown as a 60 � C and 90 � C, increasing salinity led to complete and faster release of

Table 1
Fracture cell tracer release.
Expt. Tracer eluting Tracer eluting sand location and Tracer eluting sand load/ Tracer peak Tracer peak Tracer collected at 8 FVs of
No. sand load (g) coverage area (with respect to total length of coverage (g/ft) concentration (ppm) location (FV) flow-back (% of total initial
fracture area) loading)

1 1.06 Middle, 10% 10.6 2.79 0.66 16.0%

2 0.26 Middle, 10% 2.6 0.64 0.53 17.5%
3 1.32 Inlet end, 20% 6.6 1.70 0.14 9.4%

3
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

Fig. 1. Optical micrograph of (a) uncoated sand and (b) sand coated with polymer and Rhodamine 6G.

the tracer was placed at the center of the fracture. For Experiment 3, the
peak concentration was at 0.14 FV. In this experiment, the tracer eluting
proppants were placed in the upstream 20% of the fracture; the theo­
retical peak is at 0.1 FV. This proves that the tracer peak can be used to
estimate the proppant location in the fracture. This experiment also
shows that most of tracer release happened during the shut-in period.
We observed a long tail (greater than 8 FVs) in the tracer concen­
tration curve for all three experiments. This signifies that the tracer
collected in the tail was not released during shut in, but after the start of
the flow-back. This is because the tracer release from proppants to the
fluid is governed by an equilibrium process. During shut-in, tracer is
released from proppants until an equilibrium concentration is reached.
During flow-back, tracer concentration decreases around proppants and
shifts the equilibrium so that more tracer is released, thus causing the
long tail in tracer recovery.
The amount of tracer collected in the outlet can be compared with
total amount of tracer in the fracture (calculated from the proppant
mass). 16.0%, 17.5% and 9.4% of the total tracer was collected in the
flow-back for Experiments 1, 2 and 3, respectively. The relatively low
amount of tracer recovery is partly due to tracer absorption into the
matrix (limestone in this case) during shut-in. To confirm the tracer
Fig. 2. Fraction of tracer releases as a function of time for different tempera­ absorption into the matrix, after each experiment the core was taken
ture, note that the points at 25 � C overlap with the points at 60 � C. apart and pictures were taken. The pictures of core halves in Exp. 1–3 are
shown in Fig. 6. It can be observed that parts of the limestone cores
tracer compared with the case of DI water. The time required for com­ became pink, showing rhodamine absorption into the rock matrix. Note
plete release was around 100 h and 30 h for 60 � C and 90 � C, respec­ that the pink color is most significant where the tracer-eluting proppant
tively. Note that this is significantly longer than the 5 h and 1 h for high was initially deposited, which suggests that absorption happened during
pH cases, meaning that the tracer release by high salinity is a much shut-in time in the area adjacent to tracer-eluting proppants due to the
slower process than by changing pH. During the experiment, we high tracer concentration. Also, it can be observed that tracer absorption
observed that pieces of the polymer coating were detached from the into matrix in Experiment 3 is more significant than Experiment 1 and 2.
proppants under high salinity environment. This suggests that the This explains the lower percentage of collected tracer for Experiment 3.
instability of coating under high salinity triggered the release of tracer in The absorption effect would be less in shales where the porosity and
these experiments. Furthermore, the tracer release rate was about the permeability are significantly lower than the limestone matrix used in
same for the salinity ranging from 50,000 ppm to 150,000 ppm, indi­ our experiments.
cating that 50,000 ppm is sufficient to induce tracer release. The feasibility of using tracer-eluting proppants to understand
proppants locations is demonstrated here through proof of concept
fracture cell experiments. Most of the tracer was released during the
3.5. Fracture cell tracer elution tests shut-in period and a sensible peak was detected in the flow-back curve.
The peak concentration time increased with the distance of the tracer-
We conducted three experiments with different sand loading and eluting proppants from the injection port. The fractional tracer release
location and the results are summarized in Table 1. All three experi­ will improve in the field where temperatures are higher, shut-in time is
ments were conducted with 12 ml of pH 12 buffer with 12 h of shut-in longer and tracer transport into the shale matrix is lower.
time at 25 � C. The tracer concentrations from the collected effluent for
all three experiments are plotted together in Fig. 5. To better compare 4. Numerical model for tracer release
the shape of the concentration curves, all concentrations are normalized
against the peak concentration of each experiment. A numerical model is developed for tracer transport during the flow-
For Experiments 1 and 2, the peak tracer concentration occurred back process to validate previous experimental observations. The major
between 0.5 and 0.6 fracture volume (FV), which is expected because

4
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

Fig. 3. Fraction of tracer releases as a function of time for different pH conditions at (a) 25 � C, (b) 60 � C, (c) 90 � C, (d) 125 � C.

assumptions of the model are as follows. The tracer can exist either in
cðx; 0Þ ¼ cp ðx; 0Þ ¼ c0 δðxÞ; (3)
the water (within the fracture), on the proppants or absorbed in the
shale matrix. During shut-in (soaking), the tracer is released from the �
ci �φp ci �fp φp φ �
proppants and reaches an equilibrium concentration within the water, c0 ¼ ¼ ​ ​ fp ¼ ; fm ¼ m ; (4)
φp þ φ þ φm fp þ 1 þ fm φ φ
proppants and shale matrix at the tracer loaded location (as seen in
Fig. 6). Before the flow-back, tracer can exist only at areas loaded with
fp ¼ fp n �lp : (5)
tracer-eluting proppants. The initial tracer concentration is zero else­
where. During the flow-back, tracer can be released from proppants into Boundary conditions are:
the water depending on the concentration difference between proppants
cð0; tÞ ¼ 0; (6)
and water. During the flow-back, the mass transfer in or out of the shale
matrix is considered slow and neglected.
∂2 cðL; tÞ
The mass balance equation for the tracer is: ¼ 0: (7)
∂x2
∂c ∂cp φm is a factor accounting for volumes of matrix and partitioning of
φ þ φp δðxÞ ¼ φDr2 c vφrc: (1)
∂t ∂t tracer between fracture water and matrix. fp and fm are φp and φm
The mass balance of the tracer in the proppants is modelled as: normalized with respect to φ, lp is tracer eluting proppant loading per
unit length (specified in Table 1), fnp is fp normalized with respect to lp, L
∂cp �
φp ¼ k cp c δðxÞ: (2) is the length of the system.
∂t
Substituting Eq. (2) into Eq. (1) and dividing Eq. (1) and Eq. (2) by φ
Here c is the tracer concentration in the propped fracture water, cp is gives the final governing equations:
the tracer concentration in the tracer-eluting proppants, φ is the porosity
∂c �
of the fracture, φp is a factor accounting for the polymer coating volume K cp c δðxÞ ¼ Dr2 c vrc (8)
of the proppants, D is the dispersion coefficient, v is the interstitial ve­ ∂t
locity, k is the mass transfer coefficient, is a function for tracer loaded ∂cp �
area so that δ is 1 where tracer eluting proppants are loaded and δ is fp ¼ K cp c δðxÞ (9)
∂t
0 elsewhere. c0 is the tracer concentration in the water just before flow-
back and ci is the tracer loading on tracer eluting proppants (500 ppm). where K ¼ φk : The model has a total of 4 parameters: D, K, fnp, fm. The 4
Initial conditions are: four parameters are tuned to match the tracer recovery curve of Ex­
periments 1–3. D, K, fnp were kept the same for all three experiments

5
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

Fig. 4. Fraction of tracer releases as a function of time for different salinity conditions at (a) 25 � C, (b) 60 � C, (c) 90 � C.

the tail were lower than the observed concentrations. This is likely due
to the model assumption of no mass transfer from the matrix during
flowback. Since we are using limestone to mimic the shale matrix, some
tracer can be trapped in the brine inside the limestone pores and can
subsequently flow out during flowback. This might have caused an in­
crease in the tracer concentration during flowback. For Experiment 3,
the modelled tracer concentration profile is narrower than the experi­
mental observation. This is because the current model did not consider
diffusion during shut-in. For Experiment 1 and 2, since the tracer-eluting
proppant is located away from the outlet, the tracer profile gets
dispersed as they travel to the outlet. This dispersion dominates over
diffusion during shut-in and therefore diffusion effect can be neglected.
For Experiment 3, the tracer-eluting proppant is located very close to the
outlet. Therefore, the tracer profile travels a small distance before get­
ting produced and is subject to less dispersive mixing. In comparison, the
effect of diffusion during shut-in is more significant and neglecting
diffusion leads to a narrower peak in our model (than in the experi­
ment). As the length-scale increases in the field experiments, the effect of
dispersion will further dominate over diffusion. Therefore, for field-scale
Fig. 5. Normalized tracer concentration profile for Experiment 1–3. (a) Com­ modeling, diffusion can be neglected.
plete profiles from 0 to 8 FVs, (b) Profiles from 0 to 2 FVs. The factor, fm is similar for Experiments 1 and 2. fm for Experiment 3
is higher than Experiment 1 and 2. This agrees well with the observed
while fm was allowed to change between experiments. The best matched less tracer recovery and more tracer absorption for Experiment 3. This
curves and fitted parameters are shown in Fig. 7 and Table 2, respec­ tracer release model is a simple linear model that appears to fit the
tively. The peak concentration times were well-matched and corre­ experiment results. This model can be tested in more complex fracture
sponded to the distance of tracer-eluting proppants from the core outlet. geometry and correlations can be developed between peak time and
For Experiment 1, a satisfactory match was obtained between the proppant location.
model and experimental tracer profiles. For Experiment 2, a good match
was obtained for the peak profile, but the predicted concentrations for

6
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

5. Potential for field application

The tracer eluting proppants system are very stable under low
salinity and low temperatures (up to 60 � C). This means that the tracer
eluting proppants can be easily transported and stored, whether in its
dry form or mixed with fracturing fluid before injection. As indicated in
the previous section, temperature, pH and salinity all play a role in
determining the tracer release rate from proppants. Most shale reser­
voirs are hot and formation brine salinities are high. As the proppants
are injected with cold, fresh fracturing fluid the tracer would not be
released. As the proppants settle into the fractures, the temperature of
the fluid would increase due to the subsurface temperature and the
salinity of the water would increase due to mixing with the formation
brine. These changes would trigger the release of tracer during the shut-
in period. As the fracturing fluid is produced back, the tracer would be
produced with the flow-back water. If the temperature or salinity of the
reservoir is not high enough to release the tracers, pH can be increased to
trigger tracer release. This would involve injecting an alkaline slug of
water after the proppant placement. For example, an alkaline water slug
can be injected as the last step during a fracturing job after the injection
of proppant or during drill out of plugs. Release tests should be con­
ducted with the formation brine and fracturing water mixtures at several
temperatures to identify the effective polymeric coating for the prop­
pants for a specific reservoir.
The polymer coating used in this work is sensitive to alkaline con­
ditions, but insensitive to acids. Therefore, we expect the system to
behave similarly to neutral conditions under low pH acidic environ­
ments. Since our system is compatible with different kinds of coatings,
for reservoirs with harsh acidic environments, acid-sensitive polymer
coating can be used for a pH-triggered release.
The polymer coating system in this work is unstable at high tem­
perature or high salinity conditions. If a better stability at these condi­
tions is required (for example, for operations with longer shut-in time), a
Fig. 6. Images of core halves after the elution experiments. From top to bottom:
Exp. 1, 2 and 3, respectively. Proppants were left as is in the pictures.
different polymer with better thermal and saline stability can be used.
Alternatively, multiple layers of coating can also help to increase the
thermal and saline stability.
If the tracer eluting proppants are injected towards the end of the
proppant injection (after injecting most of the regular proppants,
assuming slickwater fracturing fluid), they would likely travel the
farthest and give the estimate of the propped length (Tong and Mohanty,
2016). Currently unconventional wells often consist of multiple stages.
To identify tracer from each stage, unique tracers need to be used for
each stage. After the tracer is released from the proppants, the tracer
would flow back together with the fluid and be collected at the wellhead
in the flow-back fluid, where the tracer concentrations could be
measured. From the tracer concentration with time, the location of
proppants could be estimated for each stage. In each stage, there may be
multiple fractures. The tracer concentration would estimate the distance
travelled in each fracture. This method can be combined with fluid
tracers and microseismicity for a more accurate characterization of
hydraulic fractures.
We do acknowledge that actual fracture systems in the field will be
much more complex than our lab and simulation fracture systems. Field
Fig. 7. Comparison of tracer recovery curves for Experiment 1–3 and numer­
ical solutions. tests are needed to evaluate the efficacy of the tracer eluting proppant
technique. To scale up this technique to field application, one needs to
overcome the following concerns:
Table 2
Fitted parameters for Experiment 1-3. a Many general challenges apply to our system. For example, prop­
D ðm2 s 1
Þ K ðs 1
Þ f np ðft *g 1
Þ fm pants are injected into multiple perforation clusters during a stage.
The resulting tracer flow-back profile will be the superposition of
Exp. 1 1.55E-5 0.09 0.4 110 multiple major fracture systems. One way to address this is to inverse
Exp. 2 1.55E-5 0.09 0.4 110
Exp. 3 1.55E-5 0.09 0.4 230
model the superposed signal and try to identify the fracture system.
Another way is to calculate an effective-fracture length when sepa­
rating individual peaks proves to be difficult. If the fluid is commu­
nicating between multiple stages, then the flow-back profile will be
superposition of fractures across different stages. The tracer-eluting

7
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

proppant system would work the best when there is minimal References
communication between stages and clusters. For more complex
fracture systems, getting quantitative information out of tracer test Asadi, M., Woodroof, R.W., Malone, W.S., Shaw, D.R., 2002. Monitoring fracturing fluid
flowback with chemical tracers: a field case study. In: Presented at the SPE Annual
could be challenging. For these complex systems, we believe Technical Conference and Exhibition. Society of Petroleum Engineers. https://doi.
combining tracer-eluting proppant and existing methods may work org/10.2118/77750-MS.
the best. Bae, Y., Nishiyama, N., Fukushima, S., Koyama, H., Yasuhiro, M., Kataoka, K., 2005.
Preparation and biological characterization of polymeric micelle drug carriers with
b Fractures are long and complex systems. Actual fracture system will intracellular pH-triggered drug release Property: tumor permeability, controlled
not be perfectly linear/planar in nature; fractures may contain seg­ subcellular drug distribution, and enhanced in vivo antitumor efficacy. Bioconjugate
ments of well-propped and less-propped fractures. Our model as­ Chem. 16, 122–130. https://doi.org/10.1021/bc0498166.
Basu, S., Sharma, M.M., 2014. A new method for fracture diagnostics using low
sumes the tracer released from the proppant will stay in the propped frequency electromagnetic induction. In: Presented at the SPE Hydraulic Fracturing
section of the fracture system during flow-back. As long as the Technology Conference. Society of Petroleum Engineers. https://doi.org/10.2118/
propped segment has good continuity, this assumption is reasonable. 168606-MS.
Bhatia, K., Pande, K., 2016. First application of nonradioactive tracer technology in CSG
However, longer distance will lead to more dispersion and difficulty
unconventional basin in Central India: optimization and evaluation of fracturing
in peak identification. treatment. In: Presented at the SPE Asia Pacific Hydraulic Fracturing Conference.
c Flow in the well is not a single injection-production process. For Society of Petroleum Engineers. https://doi.org/10.2118/181782-MS.
example, after fracturing during the drill out of the plugs, some frac Catlett, R.D., Spencer, J.D., Lolon, E., Bucior, D., 2013. Evaluation of two horizontal
wells in the eagle ford using oil-based chemical tracer technology to optimize
fluid is introduced into the fracture. During this time, there is flow stimulation design. In: Presented at the SPE Hydraulic Fracturing Technology
from wellbore into the fracture. Qualitatively, this will lead to Conference. Society of Petroleum Engineers. https://doi.org/10.2118/163846-MS.
retardation of tracer peak. This could be difficult to model quanti­ Cipolla, C.L., Wright, C.A., 2002. Diagnostic techniques to understand hydraulic
fracturing: what? Why? And how? SPE Prod. Facil. 17, 23–35. https://doi.org/
tatively since additional assumptions must be made about fluid dis­ 10.2118/75359-PA.
tribution into each stage and cluster. However, the change in tracer Eisner, L., Thornton, M., Griffin, J., 2011. Challenges for microseismic monitoring. In:
peak may not be significant if the majority of tracer is released during SEG Technical Program Expanded Abstracts 2011, SEG Technical Program Expanded
Abstracts. Society of Exploration Geophysicists, pp. 1519–1523. https://doi.org/
shut-in. In other words, the effect on tracer response will be greater 10.1190/1.3627491.
for earlier stages than for later stages. In addition, for pH triggered Fisher, M.K., Warpinski, N.R., 2012. Hydraulic-fracture-height growth: real data. SPE
tracer release, pH slug could be introduced during drill out stage. Prod. Oper. 27, 8–19. https://doi.org/10.2118/145949-PA.
Gardien, C.J., Pope, G.A., Hill, A.D., 1996. Hydraulic fracture diagnosis using chemical
tracers. In: Presented at the SPE Annual Technical Conference and Exhibition.
6. Conclusions Society of Petroleum Engineers. https://doi.org/10.2118/36675-MS.
Goswick, R.A., LaRue, J.L., 2014. Utilizing oil soluble tracers to understand stimulation
efficiency along the lateral. In: Presented at the SPE Annual Technical Conference
A novel tracer-eluting proppant system is introduced for hydraulic and Exhibition. Society of Petroleum Engineers. https://doi.org/10.2118/170929-
fracturing to characterize the length of propped fractures. This tech­ MS.
Johnson, L.M., Shepherd, S.D., Rothrock, G.D., Cairns, A.J., Al-Muntasheri, G.A., 2016.
nique improves on the direct injection of tracers in frac water. This
Core/shell systems for delayed delivery of concentrated mineral acid. SPE Prod.
technique should be tested in more complex fracture geometry in lab­ Oper. 31, 351–361. https://doi.org/10.2118/173734-PA.
oratories and in the field-scale. King, G.E., Leonard, R.S., 2011. Deciphering chemical tracer results in multi-fractured
The following conclusions can be drawn from this work. Tracers can well backflow in shales: a framework for optimizing fracture design and application.
In: Presented at the SPE Hydraulic Fracturing Technology Conference. Society of
be incorporated in polymeric coatings around common proppants (e.g., Petroleum Engineers. https://doi.org/10.2118/140105-MS.
sand). The tracer release rate is a function of pH, salinity and temper­ Kumar, A., Sharma, M.M., 2018. Diagnosing fracture-wellbore connectivity using
ature. As the temperature increases, the tracer release rate increases. chemical tracer flowback data. In: Presented at the SPE/AAPG/SEG Unconventional
Resources Technology Conference, Unconventional Resources Technology
Little tracer is released below 60 � C in fresh water. As pH increases, the Conference. https://doi.org/10.15530/URTEC-2018-2902023.
tracer release rate increases. At moderate temperatures, tracer release Li, J., Pei, Y., Jiang, H., Zhao, L., Li, L., Zhou, H., Zhao, Y., Zhang, Z., 2016. Tracer
changes significantly between pH 8 and 9. As salinity increases, the flowback based fracture network characterization in hydraulic fracturing. In:
Presented at the Abu Dhabi International Petroleum Exhibition & Conference.
tracer release rate increases. Tracer release changes significantly be­ Society of Petroleum Engineers. https://doi.org/10.2118/183444-MS.
tween fresh water and 50,000 ppm salinity. The polymeric system Maxwell, S.C., Urbancic, T.I., Steinsberger, N., Zinno, R., 2002. Microseismic imaging of
should be designed for the formation brine and temperature of specific hydraulic fracture complexity in the barnett shale. In: Presented at the SPE Annual
Technical Conference and Exhibition. Society of Petroleum Engineers. https://doi.
reservoirs. The tracer concentration in the flow-back water is a function
org/10.2118/77440-MS.
of proppant position in the fracture, as shown in the single fracture core- McDaniel, R.R., Borges, J., Dakshindas, S.S., 2007. A new environmentally acceptable
scale tests. The tracer-eluting proppant position can be estimated from technique for determination of fracture height and width. In: Presented at the SPE
Annual Technical Conference and Exhibition. Society of Petroleum Engineers.
the tracer concentration profile in a single fracture. A numerical model
https://doi.org/10.2118/109969-MS.
based on convection, dispersion and mass transfer has been developed McDaniel, R.R., Holmes, D.V., Borges, J., Bajoie, B.J., Peeples, C., Gardner, R., 2009.
that matches experimental tracer concentration profiles. Determining propped fracture width from a new tracer technology. In: Presented at
the SPE Hydraulic Fracturing Technology Conference. Society of Petroleum
Engineers. https://doi.org/10.2118/119545-MS.
CRediT authorship contribution statement Muhammad, F., Guo, M., Qi, W., Sun, F., Wang, A., Guo, Y., Zhu, G., 2011. pH-triggered
controlled drug release from mesoporous silica nanoparticles via intracelluar
dissolution of ZnO nanolids. J. Am. Chem. Soc. 133, 8778–8781. https://doi.org/
Bochao Zhao: Writing - original draft. Krishna Panthi: Investiga­ 10.1021/ja200328s.
tion, Writing - original draft. Kishore K. Mohanty: Writing - original Robinson, B.A., Tester, J.W., Brown, L.F., 1988. Reservoir sizing using inert and
draft. chemically reacting tracers. SPE Form. Eval. 3, 227–234. https://doi.org/10.2118/
13147-PA.
Roussel, N.P., Agrawal, S., 2017. Introduction to poroelastic response analysis –
Acknowledgements quantifying hydraulic fracture geometry and SRV permeability from offset-well
pressure data. In: Proceedings of the 5th Unconventional Resources Technology
Conference. Presented at the Unconventional Resources Technology Conference.
The authors thank Equinor (Statoil) for the fellowship for Bochao American Association of Petroleum Geologists, Austin, Texas, USA. https://doi.org/
Zhao. 10.15530/urtec-2017-2645414.
Salman, A., Kurtoglu, B., Kazemi, H., 2014. Analysis of chemical tracer flowback in
unconventional reservoirs. In: Presented at the SPE/CSUR Unconventional Resources
Appendix A. Supplementary data Conference – Canada. Society of Petroleum Engineers. https://doi.org/10.2118/
171656-MS.
Seth, P., Manchanda, R., Kumar, A., Sharma, M., 2018. Estimating hydraulic fracture
Supplementary data to this article can be found online at https://doi. geometry by analyzing the pressure interference between fractured horizontal wells.
org/10.1016/j.petrol.2020.107048. In: SPE Annual Technical Conference and Exhibition. Presented at the SPE Annual

8
B. Zhao et al. Journal of Petroleum Science and Engineering 190 (2020) 107048

Technical Conference and Exhibition. Society of Petroleum Engineers, Dallas, Texas, Watson, W.P., Aften, C.W., Previs, D.J., 2010. Delayed-release coatings for oxidative
USA. https://doi.org/10.2118/191492-MS. breakers. In: Presented at the SPE International Symposium and Exhibition on
Shook, G.M., Pope, G.A., Asakawa, K., 2009. Determining reservoir properties and flood Formation Damage Control. Society of Petroleum Engineers. https://doi.org/
performance from tracer test analysis. Presented at the SPE Annual Technical 10.2118/127895-MS.
Conference and Exhibition. Society of Petroleum Engineers. https://doi.org/ Weng, X., Kresse, O., Cohen, C.-E., Wu, R., Gu, H., 2011. Modeling of hydraulic-fracture-
10.2118/124614-MS. network propagation in a naturally fractured formation. SPE Prod. Oper. 26,
Siepmann, F., Siepmann, J., Walther, M., MacRae, R.J., Bodmeier, R., 2008. Polymer 368–380. https://doi.org/10.2118/140253-PA.
blends for controlled release coatings. J. Contr. Release 125, 1–15. https://doi.org/ Zhang, P., Sen, M.K., Sharma, M.M., Gabelmann, J., Glowka, D., 2018. Mapping
10.1016/j.jconrel.2007.09.012. proppant distribution in hydraulic fractures in cased wellbores using low frequency
Tang, J.S., Harker, B., 1991. Interwell tracer test to determine residual oil saturation in A downhole electrical measurements. In: Presented at the SPE Hydraulic Fracturing
gas-saturated reservoir. Part I: theory and design. J. Can. Pet. Technol. 30 https:// Technology Conference and Exhibition. Society of Petroleum Engineers. https://doi.
doi.org/10.2118/91-03-08. org/10.2118/189884-MS.
Tong, S., Mohanty, K.K., 2016. Proppant transport study in fractures with intersections. Zoback, M.D., Kohli, A., Das, I., Mcclure, M.W., 2012. The importance of slow slip on
Fuel 181, 463–477. https://doi.org/10.1016/j.fuel.2016.04.144. faults during hydraulic fracturing stimulation of shale gas reservoirs. In: Presented at
Warpinski, N.R., Mayerhofer, M.J., Vincent, M.C., Cipolla, C.L., Lolon, E., 2008. the SPE Americas Unconventional Resources Conference. Society of Petroleum
Stimulating unconventional reservoirs: maximizing network growth while Engineers. https://doi.org/10.2118/155476-MS.
optimizing fracture conductivity. In: Presented at the SPE Unconventional Reservoirs
Conference. Society of Petroleum Engineers. https://doi.org/10.2118/114173-MS.