Hydraulic fracturing design improvements by proppant flowback mitigation method application during hydrocarbon production

Proppant fl owback after hydraulic fracturing treatment is quite a serious challenge that causes damage to both downhole and surface equipment, leading to unwanted workovers and production suspension, or other negative outcomes. Some special methods including preventive as well as corrective measures already exist in the industry to overcome the proppant fl owback problems. However, there is no clear guidance on how to choose the method for certain conditions, particularly during the hydraulic fracturing design stages. Therefore, the authors conducted a comprehensive review of fourteen preventive technologies to systematize them and to propose an initial decision-making approach. The following methods were considered in this paper: Resin Coated Proppants (RCP), RCP with Activator, RCP with Nanoparticles, Proppant with Liquid Resin Systems (LRS), Proppant with Surface Modifi cation Agents (SMA), Proppant with Proppant Consolidation Aid (PCA), Cylindrical Proppant, Microfi bers, Thermoset Resin Fibers, Thermoset Film Strips, Deformable Isometric Particles (DIP), Expandable Proppants, Polymer Surface Modifi ed Proppants, and In-situ Formulated Proppants. The review reveals physical, chemical, and technological properties of each method to select criteria for effective screening. So, the reader could fi nd the appropriate proppant fl owback mitigation method corresponding to a particular reservoir and the well conditions during the initial hydraulic fracturing design stages.


Introduction
Hydraulic fracturing treatments are widely considered to be a good way to produce hydrocarbons from a majority of fi elds. This is especially valid for low and ultra-low permeability reservoirs, tight oil, shale gas, mature fi elds and other unconventional reservoirs (Bomgardner 2011; Istayev et al., 2019). The main material used in the process of fracturing is a proppant to keep a fracture open after pressure is released. Conventionally, sands, ceramics, resin coated sands or ceramics, as well as bauxites are used as proppants.
A challenging problem which arises in hydraulic fracturing domain is proppant fl owback during a well's life after the treatment. Proppant fl owback results in many fi eld problems, all leading to well production decline, and some of them are listed below (Nguyen et al., 2003;: • damage to the surface and downhole equipment (pumps, screens, etc.), • premature fracture closure near the wellbore, • bottomhole and partial or full perforation blockage due to proppant settling, • frequent workovers, • back produced proppant utilization (transportation, cleaning). Actually, many methods exist to mitigate the proppant fl owback problems. They can be subdivided into preventive and post factum. Methods based on the drawdown regulation, fracture closure time regulation, injecting various bonding compositions into an existing fracture and so forth are considered as post factum, because they are implemented after hydraulic fracturing in the event that the problems occur. However, post factum methods can sometimes be used together with the preventive methods in case the latter couldn't cope with the problem itself.
On the other hand, the preventive methods are subdivided based on the pumping schedule, used materials, applied equipment and are designed in the beginning of fracturing operations. This paper considers fourteen pre- The mentioned methods are thoroughly described in literature. However, as far as is known, there is no research in terms of a methods' classifi cation and appropriateness for a particular reservoir or fi eld conditions. It is for this reason that one of the aims of this research is to conduct a deep systematization of data and information founded based on a comprehensive material review.
First of all, a special list of criteria was set based on the physical, chemical, technological, and other properties and parameters, which specify a relationship between the methods and geological and technological objects. Geological objects are those related to reservoir rocks and fl uids with related properties. Technological objects are those related to well parameters and various well intervention operations.
Furthermore, the main reasons for proppant fl owback support the decision matrix and the algorithm. That is, understating of the main reasons is one of the keys to making a decision on the application of a particular mitigation method. Based on the review of a variety of research and fi eld trials, the list in Table 1 could be considered as the common reasons for proppant fl owback.
Our systematized approach can introduce an algorithm and a matrix which could quickly and effectively  (Abbott et al., 2008;Valiullin et al., 2015;Vreeburg et al., 1994). 2 Proppant washouts Slow or fast proppant particles washouts from small cement sheaf voids, sumps, cavities near wellbore (Nguyen et al., 2003;Van Batenburg et al., 1999). 3 Partial proppant grain deformation and crash Caused by higher loads during the reservoir pressure decline, and subsequently the overburden pressure increases (Nguyen et al., 2003;Valiullin et al., 2015). 4 A slow fracture closure Resulting from slow gel breakage in the end of pumping, it can tend to the proppant settling to the bottom of the fracture with consequent rearrangements of proppant bonding agents (RCP, resin compositions). This diminishes the effect of the bonding agents (Gubanov et al., 2009;Nguyen et al., 2003). 5 Negative impact of some gel properties on the RCP bonding This is supported by Abbott et al., 2008;Gubanov et al., 2009;Nguyen et al., 2003;Vreeburg et al., 1994. For example, high pH (e.g. more than 12), persulphate based breakers, and titan crosslinkers are those leading to the degradation of RCP proppants. Gel crosslinkers could be left on the surface of RCP or on other bonding agents, which diminishes their effectiveness. 6 Increase in gas to liquid ratio Kurochkin et al., 2015 claim that this leads to an increase in proppant fl owback rate.
This could happen when bottomhole pressure is below the bubble point. The physics behind this is a multiphase fl ow which increases hydrodynamic friction in the fracture. 7 Well deviation Highly deviated wells tend to increase proppant production more than vertical ones (Browne and Wilson, 2003). 8 Produced fl uid's high viscosity Aggressive pumping schedule during Tip Screen Out (TSO), and cyclic stresses are the factors contributing to proppant fl owback as per (Letichevskiy et al., 2015). 9 Back production Up to 20% of total proppant initially pumped can be returned after a treatment (Cudney et al., 1997) 10 Weak formations Shallow reservoirs with small stresses hold proppant pack poorly. Uneven proppant distribution along the fracture with empty lags, gaps, islands of proppant are other reasons for fl owback (Lu et al., 2016). 11 Weakening of the curing process As per (Nguyen et al., 2003) in low permeability reservoirs, the fractures don't close in 24 hours, and sometimes even up to 90 days later. In such situations, when the well is put into production, RCP cannot completely cure and starts to be produced back with reservoir fl uids. Another observation is high shear stresses applied to proppant grains and coating (during pumping starting from surface and ending in the fracture). The latter is also supported by Vincent et al., 2004. 12 Gel viscosity Norman et al., 1992 mentioned that the increase in gel viscosity leads to a weakening of RCP curing. 13 Proppant size The bigger the size of proppant grains, the more stable the proppant pack . 14 Missing proppant at the top of the fracture Due to a poorly designed fracturing process with overestimated pad size and low quality gel . This tends to create channeling at the fracture top at high fl ow rates with subsequent proppant fl owback. Almond et al., 1995 also supports this idea.

The Decision Matrix for Choosing an Appropriate Mitigation Method
Based on the review of physical, chemical, and technological properties of currently available materials for application in hydraulic fracturing treatments, fourteen of the most up-to-date fl owback mitigation approaches were selected. Preventive methods are listed in Table 3. These methods are characterized by the material fi lling the fracture. Besides these, one could apply other techniques such as fl owing a well with limited drawdown, forced fracture closure, and so forth to support preventive methods. It is understandable that there are methods which can be applied in already fractured wells, for example, pumping special bonding compositions (e.g. formaldehyde resin). However, this paper discusses and considers only those methods which are included in the fracturing design stages for hydrocarbon fi elds that have potential proppant fl owback problems in their production history.
After selecting the list of potential methods, special criteria are set to reveal their practical applicability in conjunction with certain reservoir and well conditions. It is suggested to subdivide the criteria into several categories: reservoir properties, well parameters, material availability, and well intervention operations. Now each of these categories can be refi ned further into subcategories. Firstly, heterogeneity, fl uid phases and viscosities, temperature, permeability, drawdown, and rock stresses are considered as reservoir properties. Secondly, well parameters include well confi guration (vertical, deviated or horizontal) and perforation coverage. Thirdly, the material availability is subdivided as a fi eld scale and a lab scale depending on the history of its application in the industry. Fourthly, future planned well intervention operations should also be taken into account, especially workover operations with any acid and thermal treatments.
Understanding that the subject is too complex to be covered in one matrix, the following assumptions are set: • All methods are in production wells; this is a reasonable assumption because e.g. in injection wells, one of the methods is water injection by thermal proppant fracturing, which means the temperatures of the injection fl uids and the reservoir are very important. Thus, production wells are only considered. • Typical perforations are applied without screens; the assumption is based on the fact that normally screens are themselves the method to cope with particle fl owback, thus there is no need for such consideration. • All acid jobs assume only HCl or organic acids treatments, or solvent washes but without any HF Despite any applied method, a well starting time decreases after a treatment.
2 Field applicability Most of the methods were successfully tested, both in the laboratory and fi eld trials. More importantly, actual fi eld scale treatments were performed (except methods 6, 12, 13, 14, as listed in Table 3). 3 Withstanding for well fl owing rates The methods can withstand conditions of small and moderate rates, and with special constraints, high rates as well. 4 Compatibility with produced fl uids The methods can be used in majority produced fl uids.

Flexibility
The methods can be implemented in different well conditions and reservoir properties. 6 Applicability in homogeneous formations The techniques can be equally applied in homogeneous formations.
7 Fines migrations mitigation Whatever method is used, it reduces fi nes migration and movement, created during partial grain crashing. 8 Filtration properties Filtration properties of the fracture increases. 9 Durability If the method is appropriately chosen as per the criteria, it could have a long term effect. 10 Compatibility with chemicals Acid and high temperatures are two important factors to be taken into account because the composition of the majority of the material is resin, which can be affected and deteriorated by them. 11 Economic considerations Although it was stressed in the assumption list that the economic factor was purposely left out of the matrix, in case of applicability of several appropriate options by the matrix, the fi nal decision should be made by a user based on information available at the time, e.g. economics and other local conditions. Under other local conditions, the user could consider the availability in the area, logistics, policies, laws, etc. This is left for the user's own decision as a recommendation.   The comprehensive review of the studies related to RCP as well as the knowledge about the reasons for proppant fl owback revealed a set of criteria related to reservoir properties and some typical values (categorical and numerical). These criteria take into account the conditions of RCP applicability. Summarized parameters when this method is applied are given in Table 3

Proppant with Liquid Resin System
All previously discussed methods were based on proppants which are covered by resin coating in the place of origin, i.e. a plant. Another method can replace the former one by pumping a special fl uid composition during the main treatment at the fi nal pumping stages. This is called a Liquid Resin System, introduced by various authors. The generalized ranges of applicability of the method are given in the decision matrix ( The addition of SMAs to the surface of proppants allows for a decrease in proppant fl owback by fi lming the grains and thus, increasing the friction coeffi cient between them. According to laboratory experiments SMA contributes to many advantageous designs. Field observation reveals that SMAs are not suitable in high rate wells, but are more reliable in low and moderate rate wells. The summarized area of potential application of the method is given in the decision matrix in Table 3 Table 3, row 6.

Cylindrical proppant
In 2010, this method was applied for the fi rst time and tested in Arta fi eld, Egypt. Technologically, this proppant could be applied in almost any situations and conditions, and the only constraint for the method is its cost and economics, as well as its availability in a particular region. Letichevskiy et al., 2015 list the area of cylindrical proppant applications. Cylindrical proppant is normally made of bauxite, which means two consequences: fi rst, it has high strength, and second, it costs more than RCP. Some authors concluded that the method allowed for the production of more hydrocarbons in comparison with conventional RCP application as per the results of total production from wells. The details are in Thermoplastic Film Strips technology was fi rst introduced in the laboratory and fi eld research of Nguyen et al., 1996. The technology implies the addition of TFS into proppant when mixing. TFS is a material represented as strips with a certain length (normally the length of a strip is the length of several grains), and it can bond proppant by increasing the friction factor between grains. Summarized conditions of TFS application are provided in Table 3

Expandable proppant
One of the newest laboratory proved technologies is Expandable Proppants. This technology was not tested at a fi eld scale, but nevertheless, it has a great potential in dealing with proppant fl owback. The material is made of shape memory polymers (SMP) which start increasing their volumes under a certain downhole temperature, thus packing the fracture. Based on the information from Santos et al., 2018 studies, we defi ned the applicability criteria for the methods and summarized in the decision Table 3, row 12.

Polymer Surface Modifi ed Proppant
In the study of Fu et al., 2016, a new material to deal with proppant fl owback was introduced. The material coats the proppant grain surfaces. The composition of the material is poly 2-fl uorine-4vinilpyridine with the addition of methanol. The proppant grains' surface modifi cation allows for a proppant pack to re-aggregate after disaggregation. For the moment, this method was only laboratory-scale tested without any fi eld trials. The main conditions when the method can be applied are listed in the decision matrix in Table 3, row 13.

Fu et al., 2016.
14 In-situ Formulated Proppant One of the latest inventions is a material that creates a hard proppant from a liquid at downhole conditions. The technology was introduced by Chang et al., 2015 as an alternative to conventional proppants. The idea of the method is to formulate hard grains from liquid compositions when the latter enters the fracture. The grains can withstand high stresses up to 95 MPa and no damage or creation of fi nes occurs. The big advantage of the technique is that it is pumped in liquid state, thus no risk with premature screen-out is present.
In Table 3, row 14 the main conditions when the method is applicable are given. due to many uncertainties in economic factors and the complexities in business choices, the economic factor was intentionally left out from the decision matrix. The next step is to create a special matrix which consists of "0" (zeros) meaning a method's inappropriateness and "1" (ones) -the appropriateness for the given conditions (see Table 3). In addition, the matrix has a "green-yellow-red" colour code. Green with "1" meaning full appropriateness, yellow (with "1" or "0") -some cautions and uncertainties, and red with "0" -full rejection or withdrawal controversial operation (e.g. refuse HCl or stream treatments). Further clarifying, "1" and yellow represents a situation when the method could be applied in the condition, but additional considerations are required. "0" and yellow could be considered as applicable if special additional conditions are provided. The appropriate conditions are different for various methods, thus a reader should carefully pay attention to that particular case and it is recommended to fi nd more in related supportive references in Table 4.

Chang et al., 2015.
Through the application of the above rules of the matrix, a user can easily do a quick look analysis and screening using the matrix in order to make a fi rst decision on proppant fl owback mitigation method selection corresponding to a particular reservoir and the well conditions. This is based on the fundamental baseline supported by analysis, research, and systematization of more than fi fty papers, books, and sources of information which provide the results of laboratory experiments, fi eld observations, analytical, statistical, and other stud-ies. Such categorical quick fi ltering could have some limitations due to situations where not all parameters are provided with certain quantitative ranges. For example, one cannot exactly say what is the maximum or minimum stress when we set high, intermediate or low stresses. This is for the reader's decision for particular given cases in a location. The same is applied for permeabilities, drawdowns, viscosities, and deviation.
Setting aside particular characteristics of each mitigation method, some general common features for all methods were revealed and are represented in Table 2. All the supporting references for this generalization are given in Table 3 for each method.
Further in the paper, the ground basis of the matrix applicability is provided, being supported by given conditions and certain distinctive characteristics of the methods.

The Decision Matrix Basis
Based on the comprehensive review of a variety of research on the subject, the following classifi cation descriptive table was generated (see Table 4). The classification table provides the decision matrix shown in Table  3 with the references where one or the other "0" and "1" with colour codes are supported. Such a method of presenting the matrix allows for the covering of a lot of information from different sources in a convenient and effi cient manner.

An Algorithm to Select a Proper Method to Mitigate Proppant Flowback Problem
Based on the decision matrix provided in Table 3, a new algorithm is introduced that provides a probable initial decision about what method to apply on a particular given fi eld (see Figure 1). It is assumed that every case needs its special attention and additional conditions related to a specifi c method under consideration, but they should be considered as secondary factors and fi lters. Eventually, the fi nal decision criteria are economics and logistics which are not considered in our study due to high uncertainties in the subjects.
Apart from many advantages of proposing the algorithm and decision matrix, some other limitations are present that also have to be taken into account. For example, some criteria are not set numerically, but set qualitatively in the form of some category, e.g. while screening relative to permeabilities, stresses, viscosities, drawdowns we don't provide exact ranges of applicability in digital form for any of the methods. That's why such qualitative classifi cation contains some uncertainty that an engineer should keep in mind during decision making. Thus, only relative comparisons could be implemented. Nevertheless, even under such limitations, our proposed algorithm and decision matrix could give the initial ranking of the methods relative to a given problem, provide a quick look analysis, which is normally far cheaper than the application of numerical simulations or providing experimental lab or fi eld trials. Moreover, such a quick look analysis could support what method to apply in the laboratory fi rst, and after lab analysis a fi nal decision for fi eld trials could be made.

Conclusions
In this paper, the decision matrix is proposed to create a quick look selection of an optimum proppant fl owback mitigation method. The initial decision schemes could help fracturing and production engineering teams during the proppant hydraulic fracturing design stages, and provide the basis for a good choice.
Through the review and systematization of fourteen up-to-date preventive methods, the following conclusions can be drawn: 1. To facilitate a comprehensive review, the following selection criteria were set: a. Proppant fl owback reason identifi cation, b. Physical, chemical, and technological properties of proppant materials, c. Geological and other formation properties, d. Well parameters, e. Technological capabilities, f. Various operational processes. 2. Common reasons for proppant fl owback were identifi ed and systematized.
3. Setting aside particular characteristics of each mitigation method, some general common features for all methods were revealed.
4. Based on the criteria, classifi cation, description, and systematization, the decision matrix was tabulated allowing for the initial ranking and decision-making on what proppant fl owback mitigation method is the best one to apply under certain fi eld conditions. The decision matrix could help in advance to fi nd a method with some accuracy and assumptions in hydraulic fracturing design stages. In doing so, convenient ways of data and condition representation by zeros (0) and ones (1) with colour codes supply the matrix. 5. A special algorithm on how to work with the decision matrix is developed. The algorithm is represented in the form of a fl owchart facilitating in the decision-making process.
6. The proposed algorithm and decision matrix could provide a quick look initial selection to support numerical simulations or experimental lab or fi eld trials.