Introduction

Carbonate reservoirs occupy an extremely important position in the world’s oil/gas resources. And more than 30% of carbonate reservoirs are fractured-cavity reservoirs (Perez Rosales et al. 2002). The exploration and development of carbonate reservoirs in China present a rapid development trend as the increasing demand for energy resources (Chen et al. 2005). Due to the extremely low permeability of carbonate rock matrix (between 0.1 and 1 mD), the vugs and fractures are the fluid flow spaces. The flow pattern in the large vugs is recognized as pipe-flow mode which is much different with seepage in porous media. The flow pattern in fractures is sometime near to seepage in porous media, while sometime near to pipe-flow mode. Therefore, we cannot simply use the conventional parameter (permeability) to characterize the fractured-vuggy reservoir because of its complicated storage spaces (Zheng et al. 2010; Du et al. 2011; Xu et al. 2010). Strong heterogeneity, complex fracture-cavity structure, and connections are the main characteristics of fractured-cavity carbonate reservoir (Cruz Hernandez et al. 2001). Moreover, the main storage space and seepage path possess the features of discontinuity and randomness (Popov et al. 2009; Lu et al. 2010; Chen et al. 2010). After years of water-flooding development, many shortcomings such as short period of stable production, early water breakthrough, rapid increase of water cut, and high natural declining rate are exposed gradually.

Since current conventional recovery methods are no longer feasible, substantial portions of crude oil have remained uncovered. With the help of enhanced oil recovery (EOR) methods, which are part of the more general scheme of improved oil recovery (IOR), much of this oil could be recovery. However, the choice of the method and the expected ultimate recovery is dependent upon many considerations and economic factors, as well as technological factors (Firouz and Torabi 2014). The solvent gases can be carbon dioxide (CO2), flue gas, and light hydrocarbon gases such as natural gas and methane. Cyclic solvent injection is a single-well EOR method which was initially considered to be an alternative technology to improve oil recovery. It is performed by injecting gas into a well (huff cycle), followed by a relatively short shut-in time, and then, the well is returned to production after a soaking time to allow for solvent interaction with the formation oil (puff cycle). Numerous laboratory and field tests have been performed, and the results reveal that the process is economically feasible (Qazvini Firouz 2011; Palmer et al. 1986; Simpson 1988; Haskin and Alston 1989; Alfarge et al. 2017; Cudjoe et al. 2016). However, most of the works reported, either in the laboratory or field trials, have been focused on the conventional single-porosity reservoirs (Torabi and Asghari 2010).

One of the first attempts to investigate the efficiency of the huff-and-puff technique on viscous oil was conducted by Schnitzler et al. They performed an experimental study on a laboratory-scale core having the dimensions of 3.8-cm diameter and 22.8-cm length. Based on their results, the ultimate recovery factor (URF) was increased by 14% during 2 cycles of CO2 injection. A. Kantzas et al. (1993) designed a gravity-assisted immiscible cyclic gas (N2) injection experiment applied in fractured-cavity rocks. The experiment was on the impact of heterogeneity and wettability on oil recovery using a two-dimensional micro-glass tube model. The experimental results showed that the gas displacement could increase the recoverable reserves of crude oil by 20% even though in strong heterogeneous and weak hydrophilic reservoirs. carried on a high-temperature emptying experiment with fractured-cavity rocks to investigate the mechanisms of cyclic gas injection. The study found that the N2 EOR mechanisms include thermal expansion, thermal enhancing dissolved gas driving, dry distillation, and steam driving. Sanchez Bujanos et al. (2005) analyzed the production performance of N2 injection in complex reservoir through the pressure monitoring system. The production and pressure data indicated that N2 was isolated within the gas cap and complement energy for the formation. Zhao et al. (2010) designed a cavity-porosity core to research the impact of viscosity and angle on oil recovery and the effects of N2 and CO2 on oil/water and gas/water interface. In CO2 injection, the effect of angle on oil recovery became more sensitive, and gravity differentiation was more important. In N2 injection, it took more time to form the gas cap with the increase of pressure. CO2 could efficiently alter the location of oil/water interface when there were the vertical fractures in fractured-cavity reservoirs. And N2 gas cap could keep on the high-pressure production formation due to its low solubility in water and oil. Liu et al. (2012) and Yuan et al. (2015) designed and fabricated a visualized fractured-cavity model based on fracture-cavity relations of Tahe fourth block. The experiments indicated that there existed much remaining oil after bottom water breakthrough, and determined the five main types of the remaining oil. Li et al. (2008) researched the feasibility of N2 injection on developing the attic oil of fractured-cavity reservoirs and analyzed its main mechanism including gravity differentiation, oil expansion dissolved gas, and altering the flow direction. They concluded that N2 injection on improving oil recovery for single wells was feasible. Wang (2009) studied the flow mechanism of three phases (oil/gas/water) in fracture-cavity mediums combining the numerical simulation and theoretical analysis, which had provided the theoretical foundation on the gas injection for EOR.

This paper presents the results of an experimental study on the feasibility of gas medium huff-n-puff in the single well set up in a 3D physical simulation model. The influence and mechanisms of gas injection position were studied through 2D visualized model. Then two groups of N2 huff-n-puff experiments were conducted in the three-dimensional resisted model to demonstrate the results concluded through the experiments conducted in the 2D model.

Methodology

Experimental physical models

A 2D visualized physical model and a 3D pressure resistant model were designed and fabricated according to the similar theories, which have been described in detail in previous documents (Li et al. 2014; Hou et al. 2014). Tahe oilfield in Tarim Basin of China is a typical fractured-vuggy carbonate reservoir. Large karst vugs, dissolved fracture-vugs, and multi-scale fractures are the main storage and flow spaces. Matrix does not have storage and permeability capabilities. Therefore, the flow pattern in the large vugs is recognized as pipe-flow mode which is much different with seepage in porous media. The flow pattern in fractures is sometime near to seepage in porous media, while sometime near to pipe-flow mode (Lyu et al. 2017; Hou et al. 2018). Therefore, we cannot simply use the conventional parameter (permeability) to characterize the fractured-vuggy reservoir because of its complicated storage spaces.

The 3D physical simulation model was designed and fabricated according to the research on geological model of S48 unit in Tahe oilfield; well group TK467-S48 was selected as the prototype for model design. Longitudinal sections of the geological prototype were divided into six layers. And carbonate powder and epoxy resin were used to suppress the six layers with the same diameter of 40 cm and a height of 5.6 cm. Caves and fractures were carved based on the geological prototype as were shown in Fig. 1. Six layers were bonded together and put into the cylindrical mold, encapsulated, and stereotyped with epoxy resin. Figure 2 shows the well locations and the surrounding fracture-cavity structures. Three wells were designed in the model based on geological wells (TK467 and S48). The red dotted portion in Fig. 2b was filled with sands, so the filling degree of well TK467 and S48 were fully-filling and half-filling. The wells’ type and depth were shown in Table 1. The total volume of fractures and caves of the physical model is 3313 cm3, among which the filling sands occupies 450 cm3. The filling degree is 13.6%.

Fig. 1
figure 1

Comparison of the fractured-cavity structures between the physical model and the real core for each layer

Full size image
Fig. 2
figure 2

Well locations and the surrounding fractured-cavity connection structure

Full size image
Table 1 Well parameters of the 3D physical model
Full size table

The 2D visualized model was designed based on the stratigraphic profile of well S48 and well TK467 in fourth block Ordovician fractured-cavity carbonate reservoir of Tahe oilfield, as shown in Fig. 3. Small-scale fractures and cavities were etched on the artificial carbonate core. The model was filled with sand and sealed by transparent epoxy resin. The size of the model is 30 cm × 30 cm × 5 cm, and entire volume of fractures and cavities is 345 mL, and filling degree is 30%.

Fig. 3
figure 3

Fracture-cavity connection of S48 unit (a) and the real figure of macroscopic visualized model (b)

Full size image

Experimental materials

An oil sample representing the crude oil in Tahe oilfield, with density of 0.853 g/cm3 and a dead oil viscosity of 24.7 cp at 65 °C, was used to saturate the 3D physical model. And another oil sample with the same viscosity at 25 °C used in two-dimensional visualized experiments was prepared with liquid paraffin and kerosene. The oil composition was shown in Table 2. The brine phase in all experiments is the synthetic brine according to the reservoir water analysis results of well S48 in Tahe oilfield, and the salinity is 20 × 104mg/L (Mg2+ 1440 mg/L, Ca2+ 12,000 mg/L, Cl− 132,000 mg/L, \( {\mathrm{SO}}_4^{2-} \) 580 mg/L, \( {\mathrm{HCO}}_3^{-} \) 105 mg/L). Carbon dioxide and nitrogen with stated purities of 99.99 and 99.97%, respectively, were used. And the gas mixture of CO2 and N2, volume ratio of which is 1:1, was pressured into a container. The carbonate powder and epoxy resin were used to fabricate the 2D visualized model and 3D physical model.

Table 2 Composition of crude oil of well TK467 in Tahe oilfield
Full size table

Experimental setup

Figure 4 shows the schematic of the huff-n-puff experimental setup in 3D physical model. It consists of the injection system, oil production system, and temperature controller system. In production system, a back pressure regulator and a gas-liquid separator were utilized to control the back pressure and measure oil/gas production rate. Besides, three highly sensitive digital pressure gauges, directly connected to the three production wells, were used to record the pressure throughout each experiment. It was also used to observe the pressure depletion during the soaking time, which is an indication of the diffusion process and the solubility of the gas in the oil phase and the fractured-cavity system. In addition, a separate vacuum setup was used to vacuum and saturate the model prior to each experiment.

Fig. 4
figure 4

The schematic diagram of gas huff-n-puff experiments in 3D physical model

Full size image

The schematic diagram of the experimental setup used in 2D visualized huff-n-puff study is shown in Fig. 5. The main components of this setup are as the following:

  1. (1)

    The injection stream: the main components of an injection stream are the injection line, a N2 cylinder, a CO2 cylinder, a gas mixture cylinder, two piston-cylinder apparatuses full of the crude oil and brine, and a high-pressure syringe pump (Teledyne ISCO). Besides, a gas flowmeter in inlet was used to control the gas injecting rate.

  2. (2)

    Oil production stream: consists of a production line and production valve connected to the production well.

  3. (3)

    The video system: a light video camera (distinguishability: 1920 × 1080) was used to record the phenomenon of oil/water interface change, the remaining oil distribution, and gas huff-n-puff to start the remaining oil. And a LED panel light was used to ensure the image quality.

  4. (4)

    Temperature controller system: to maintain a constant temperature of 65 °C and to ensure that the temperature inside the air bath is uniform. A thermo-tank including one heater, two fans, and a digital temperature controlled was used.

Fig. 5
figure 5

The schematic diagram of 2D visualized huff-n-puff experiments

Full size image

Experimental procedures

In this paper, two kinds of experiments were conducted in 3D model: (a) the huff-n-puff experiments in the single well S48 to study the dynamic performance of three kinds of gas mediums including N2, CO2, gas mixture and (b) pure N2 huff-n-puff experiments in two wells of TK467 and S48 to investigate the impact of gas injection position on the puff performance and EOR effect. The specific procedures are as follows:

  1. (1)

    All the equipment was firstly set up according to Fig. 4.

  2. (2)

    Vacuumize the 3D model, and then saturate the synthetic brine and simulated oil to calculate the original oil saturation and irreducible water saturation; finally, set up the back pressure regulator to 3 MPa.

  3. (3)

    Bottom water displacement: connect “water gravity installation” to the bottom water well and open the well S48 at the same time to simulate the depletion bottom water driving. When the water content reached to 98%, well S48 was then transferred to gas injection well.

  4. (4)

    The pressure of gas container, physical model, and back pressure was set up 3 MPa, and the gas medium was injected into well S48 at the speed of 5 mL/min. The total amount was 100 mL at the condition of 3 MPa and 65 °C.

  5. (5)

    Puff phase: well S48 was shut in for 24 h (soaking time) after gas medium was injected. Then the puff stage started. The oil/gas/water production rate and pressure were recorded.

Three kinds of experiments were conducted in 2D visualized model: (a) bottom water flooding; (b) oil recovery mechanisms analysis of N2, CO2, and N2/CO2 mixture injected in well TK467; and (c) oil recovery comparison of N2 huff-n-puff in low (TK467) and high positional wells (S48). The pressure of back pressure was set up 1 MPa, and the gas medium was injected into well TK467 at the speed of 2 mL/min. The total amount was 25 mL (0.1 PV) at the condition of 1 MPa and 65 °C. The aim of bottom water flooding experiment is to investigate oil/water flowing mechanism, formation mechanism, and distribution of residual oil, which was fundamental to analyze the oil recovery mechanisms of N2, CO2, and N2/CO2 mixture. The influence of gas injection position on oil recovery in N2 huff-n-puff process was then investigated.

Results and discussion

Oil/water/gas production analysis of N2/CO2 huff-n-puff experiments in the 3D model

The single well S48 of the 3D model was selected to study the dynamic performance of three kinds of gas mediums including N2, CO2, and gas mixture during huff-and-puff processes. The other well TK467 was shut in. Meanwhile, the parameters of gas/oil ratio, oil production rate, and gas production rate in the puff stage were analyzed and compared.

Oil/water production performance during the huff-and-puff process

The huff-n-puff experimental production curves of N2, CO2, and N2/CO2 mixture in the single well S48 were shown in Figs. 6, 7, and 8. The experimental procedures of each gas medium huff-n-puff were divided into three stages including bottom water displacement phase, soaking phase, and puff phase. At the beginning of the bottom water displacing phase, the well S48 produced oil stably and there was no water produced. At the time of 178 min, the bottom water occurred and the water cut suddenly reached 100%, which showed eruptible water breakthrough. And it could be inferred that the oil/water interface went up horizontally before 178 min. The process of bottom water flooding was considered as piston displacement. The water occurred breakthrough within 20 min when the oil/water interface reached in the bottom of well S48. Finally, 9.8% OOIP was recovered in bottom water driving phase.

Fig. 6
figure 6

Dynamic production performance of N2 puff-and-puff

Full size image
Fig. 7
figure 7

Dynamic production performance of CO2 puff-and-puff

Full size image
Fig. 8
figure 8

Dynamic production performance of gas mixture of N2/CO2 puff-and-puff

Full size image

When the water cut of well S48 reached 100%, the well S48 was shut in and the bottom water was continued injected. Afterwards, the model pressure was quickly rose to 3 MPa. The injected water was to pressurize the model according to its extremely low compression coefficient. At this moment, the nitrogen container, the 3D model, and the back pressure regulator were all set to 3 MPa. Then the 100 mL nitrogen at the condition of 3 MPa and 65 °C was injected into the model. The soaking time started. The model pressure gradually rose and reached to the stable state within 10 h. The model pressure increased by 0.207 MPa and reached to 3.207 MPa finally in the soaking phase. Because pressure gradient along the model is very small, the flow process is primarily dominated by molecular diffusion and capillary forces (Hua et al. 1991). The injected gas volume expanded due to thermal effect. The pressure change after nitrogen injection reflected its effect on stratum energy complement. And the nitrogen gas could accumulate on the top of the model and form gas cap due to insolubility and oil/gas density variation. When the model pressure kept stably, the well and bottom water were both opened and the back pressure operator was set to 0 MPa. Overall, the puff phase was a short process and the pressure decreased to 1.488 MPa within 20 min. The water cut decreased rapidly to a stable state of average 68.5% which was due to the gas cap energy inhibiting bottom water coning. When the gas cap energy was less than that of bottom water, the water cut and the model pressure increased again with the injection of bottom water. When the gas cap energy was much more than that of bottom water, the water cut decreased. Therefore, both the pressure and the water cut presented the state of fluctuating. The puff phase lasted for 120 min and the final oil recovery during puff stage increased by 35.53%. The fluctuating curve of model pressure and water cut proved nitrogen functions of gravitational differentiation, water coning controlling, and stratum energy complement.

The dynamic production curve of CO2 huff-n-puff in well S48 was shown in Fig. 7. The experimental procedures were similar to N2 huff-n-puff. Compared with that of N2 huff-n-puff, the model pressure in soaking phase decreased gradually a little bit to 2.96 MPa. Molecular diffusion contributes to enhanced oil recovery in puff period (Jia et al. 2018). In this study, molecular diffusion contributed to 0.2–0.3% of oil recovery factor. It could be inferred that the oil viscosity remain unchanged and viscosity reduction with CO2 dissolution was not apparent. The apparent oil viscosity reduction after CO2 dissolved reflected the remarkable pressure decrease. Adversely, the slight soaking pressure change reflected tiny oil viscosity reduction. In puff stage, the model pressure decreased to 0 quickly, and it did not show the fluctuated characteristic. Besides, the water cut could merely decrease to an average value of 92.5% and the oil recovery during puff stage improved only by 4.78%, which was far less than that of N2 huff-n-puff. Therefore, the effect of CO2 dissolving in oil and reducing oil viscosity was taking the second place while the energy complement of inert gas was in dominant position.

Given the limitation of the single gas, the gas mixture of N2 and CO2 was injected simultaneously to investigate whether the synergistic effect existed. The experimental result was shown in Fig. 8. The model pressure also gradually increased by 0.128 MPa, which was less than the single nitrogen. The water cut decreased to an average value of 80% and the oil recovery during puff stage increased by 17.85%, both of which were less than single N2 huff-n-puff while much more than the single CO2 huff-n-puff. Overall, the nitrogen in the gas mixture played a major role in reducing water cut and improving oil recovery.

Gas/oil production rate and gas/oil ratio

The experimental production parameters including gas/oil ratio, oil production rate, and gas production rate in the puff stage were analyzed and compared of N2, CO2, and N2/CO2. The three parameters could reflect and prove that the nitrogen gas was more suitable to improve oil recovery for light oil in fractured-cavity reservoirs. The production parameters of N2, CO2, and N2/CO2 were shown in Figs. 9, 10, and 11. The production process in N2 puff stage could be divided into three phases. The period of first 20 min was the early stage in puff process, the features of which were that the gas and oil production rate reached to 100 mL/min rapidly, but the gas/oil ratio maintained around 15. The produced oil in the first stage accounted for 56.2%. The first stage was similar to the energy failure stage and huge quantity of the crude oil was produced in a short period time. The second stage was in relative stable period in which the oil and gas production rate was low. But it maintained a long production time when 32.3% of the total improved oil in the whole puff stage was produced. The third stage was the later and short period in which the gas/oil ratio increased and only 11.5% of oil was produced. It could be inferred that the gas/oil interface reached to the bottom of well S48 and the gas breakthrough was formed.

Fig. 9
figure 9

Oil/gas production rate and gas-oil ratio of N2 huff-and-puff

Full size image
Fig. 10
figure 10

Oil/gas production rate and gas-oil ratio of CO2 huff-and-puff

Full size image
Fig. 11
figure 11

Oil/gas production rate and gas-oil ratio of N2/CO2 mixture huff-and-puff

Full size image

As was shown in Fig. 10, the puff process of carbon dioxide was divided into two stages. Of the oil, 75% was produced in the first stage in which pressure was released immediately. The dissolved gas overflowed and carried huge quantities of crude oil. The production mode in first stage was similar to energy failure production and solution gas driving. And the average production rate of oil and gas kept relative high (around 50 mL/min). While the oil and gas production rate declined sharply in the second stage. The gas/oil ratio increased gradually and kept relative stable, which inferred that a small amount of dissolved gas played a role of the dissolved gas driving in the second stage. Only 25% of the oil was produced in this stage. The puff process of the gas mixture was similar to that of pure carbon dioxide. The gas ratio in the second stage changed a lot, which was different from the stable state of pure carbon dioxide, which indicated the mechanism of energy accumulation and release due to the mixed nitrogen. The oil recovery of gas mixture was much more than that of pure carbon dioxide, but less than that of pure nitrogen, as was shown in Fig. 11.

Therefore, it could be concluded that nitrogen for energy complement and carbon dioxide for dissolved gas driving were the two mechanisms contributed to enhanced oil recovery. Here, energy complement, accumulation, and release triggered by nitrogen played a dominant role. Whether the cap energy could be formed play a decisive role on the effect of improved oil recovery in fractured-vuggy reservoirs, which was different from the conventional sandstone reservoir that the dissolved gas driving triggered by carbon dioxide may be the main mechanism.

Remaining oil starting analysis of N2/CO2 huff-n-puff in the 2D model

The huff-n-puff experiments in the fractured-cavity 3D model illustrated that the gravitational differentiation of N2 played a dominant role in improving light oil recovery. Three groups of huff-n-puff experiments were conducted in the designed 2D model to study the oil recovery mechanisms of N2, CO2, and N2/CO2 mixture. The remaining oil type and distribution after bottom water flooding were firstly analyzed.

Distribution of remaining oil after bottom water flooding in 2D model

Oil/water interface change with bottom water flooding was shown in Fig. 12, from which the residual oil formation and distribution can be observed clearly. At the beginning of bottom water flooding, the oil/water interface went up horizontally when the bottom water was injected. The process of bottom water flooding was considered as piston displacement and pressured pipe flow (Fig. 12b). With the injection of bottom water, the oil/water interface was elevated to step up. When it was higher than the narrow cave in the middle side, the oil could not be displaced. The reason is due to the complex shape of caves and water flow deviated from the breakthrough channel (Fig. 12c). As a result, different types of the remaining oil were formed and distributed in different position (Fig. 12d). The formation mechanisms of the remaining oil are poor connectivity between caves which bottom water could not sweep. The types of remaining oil include those trapped in the closed fractures and caves (A), bypass oil (B), attic oil (C), and oil film absorbed on the surface of fractures (D). Remaining oil was trapped in the closed caves due to bad connectivity between fractures and caves. Because the minimum resistance was the fluid flow direction as was shown by blue arrow in Fig. 12, the bypass oils were easy to form due to the poor swept volume. The attic oils were trapped in the top of caves because of the oil/water density difference. In the caves filled with sands, the oil films formed and affected by wettability, oil viscosity, and temperature.

Fig. 12
figure 12

Oil/water distribution during bottom water flooding process

Full size image

Remaining oil recovery comparison of N2, CO2, and N2/CO2 mixture

The oil recovery principle of N2, CO2, and N2/CO2 mixture in puff stage was in Figs. 13, 14, and 15. Of the nitrogen gas, 0.1 pore volume was injected under the condition of 1 MPa and 65 °C. The soaking time ran for 1 h and the well TK467 was then changed into the production well. As was shown in Fig. 13b, the attic oil in the bottom cave was replaced as the majority of the injected nitrogen gas stored in the bottom part in the initial puff stage. Then nitrogen transferred to the cave in the upper part due to the density difference, and the gas/oil interface appeared gradually (Fig. 13c). The attic oil in the upper part and the bypass oil were totally activated (Fig. 13d). In CO2 puff stage, the attic oil in the bottom cave was firstly activated as CO2 gas stored and dissolved in the bottom part (Fig. 14b). With the puff went on, the gas/oil interface did not appear in the upper part and the gas breakthrough occurred in the production well TH467. The reason was that the majority of CO2 dissolved in the bottom water and CO2 could not move to caves in the upper part. Only small part of the upper attic oil and bypass oil were recovered (Fig. 14d). The oil recovery process of N2/CO2 mixture was shown in Fig. 15. In the puff stage, part of the nitrogen gas transferred to the upper part and replaced the attic oil and bypass oil. Part of CO2 dissolved in the bottom water in the bottom cave, which was not beneficial to improve oil recovery. On the whole, the oil recovery effect of N2/CO2 mixture was close to that of N2, but much better than the pure CO2.

Fig. 13
figure 13

Remaining oil startup principle during N2 huff-n-puff in well TK467

Full size image
Fig. 14
figure 14

Remaining oil startup principle during CO2 huff-n-puff in well TK467

Full size image
Fig. 15
figure 15

Remaining oil startup principle during N2/CO2 huff-n-puff in well TK467

Full size image

Effect of injection position on N2 huff-n-puff

The 2D visualized model was designed to analyze the remaining oil distribution after bottom water flooding, in which two groups of N2 huff-n-puff experiments were conducted in high-position well S48 and low-position well TK467 to study the effect of gas injection position on the remaining oil activity. Then two groups of N2 huff-n-puff experiments were conducted in the 3D resisted model to demonstrate the results concluded through the experiments conducted in the 2D model.

Effect of N2 huff-n-puff with different injection position in 2D model

The production performance of N2 huff-n-puff in high-position well S48 in 2D model was shown in Fig. 16. In the process of gas injection, nitrogen tended to flow along the blue direction and accumulate on the top of the middle cave due to gravitational differentiation (Fig. 16b). But it could not flow into the caves in left side which were filled with sands. Thus, the resistant was much larger than the caves in the right side. In puff stage, the attic oil in the middle cave was recovered due to the effect of gas cap, as the red cycle showed in Fig. 16c. But it was easy to form gas breakthrough along the blue line to well S48 when the gas/oil interface reached the bottom of well S48 (Fig. 16d). Besides, part of the remaining oil was driven to the cave in the lower side and formed the new type of remaining oil, as the yellow cycle showed in Fig. 16e.

Fig. 16
figure 16

Remaining oil startup principle of N2 huff-n-puff in high positional well S48

Full size image

Similarly, the production process of N2 huff-n-puff in single well TK467 was shown in Fig. 17. Nitrogen tended to accumulated on the top part of right-side caves in the process of gas injection. In puff stage, the gas/oil interface went down slowly and replaced the remaining oil, the displacing process of which was like the piston flooding. The gas was not formed breakthrough channel until the oil/gas interface reached the bottom of well TK467, as was shown in Fig. 17e. All types of the remaining oil in the right side were recovered. As a result, N2 huff-n-puff should be implemented in the low-position well in case of the early gas breakthrough.

Fig. 17
figure 17

Remaining oil startup principle of N2 huff-n-puff in low positional well TK467

Full size image

Oil production volume with different N2 injection position in 3D model

N2 huff-n-puff experiments in the single well S48 and well TK467 were conducted, respectively, in 3D physical model. The experimental condition and results were shown in Table 3. Overall, the oil recovery during nitrogen puff phase in the low-position well TK467 was 43.25%, while it was 35.53% in the high-position well S48. The results were consistent with that observed and concluded in the 2D visualized model. As was shown in Fig. 18, the gas production rate during puff stage in well S48 was much higher than that in well TK467. It reflected that the nitrogen gas was easy to form gas channeling if it was injected from high-position well, which resulted in relative lower gas utilization.

Table 3 Comparison of N2 huff-n-puff oil recovery effect in different position
Full size table
Fig. 18
figure 18

Gas production rate during puff period of well S48 and well TK467

Full size image

Conclusions

From this study, the following conclusions can be drawn:

  1. (1)

    The designed and fabricated 2D and 3D models are closer to the reservoir situation of fourth block, Tahe oilfield. The production performance of gas huff-n-puff experiments could represent and instruct the real production condition.

  2. (2)

    The effect of CO2 dissolving in oil and reducing oil viscosity is taking the second place while the energy complement of N2 is in dominant position. In the gas mixture, the nitrogen played a major role in reducing water cut and complementing formation energy. The nitrogen gas gradually migrates to the high structural position and forms gas cap for energy supplement. It replaces the attic oil in the top part and bypass oil in the bottom part, and redistributes the remaining oil.

  3. (3)

    It could be concluded that there exist four types of the remaining oil including those trapped in the closed fractures and caves, bypass oil, attic oil, and oil film absorbed on the surface of fractures. All types of the remaining oil could be recovered if N2 huff-n-puff measure is implemented in the low-position well in case of the early gas breakthrough.

  4. (4)

    The 3D model with complicated fracture-vug structure could represent the fracture-vuggy reservoir to some extent. But the experimental pressure is not easy to reach the field condition (60 MPa), so we still need to improve the model both in the structure and in the pressure condition. Besides, the influencing parameters of soaking time, huff-n-puff cycles, production rate, and so on should be evaluated in the future work.