J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 245 Total Pages: 11 Nodal Analysis for Unconventional Reservoirs—Principles and Application Wentao Zhou, Raj Banerjee, and Eduardo Proano, Schlumberger Summary Nodal analysis is the standard technique used to evaluate the performance of integrated production systems. Two curves represent the capacities of the inflow and of the outflow, and the intersection of the two curves gives the solution operating point. Limitations of traditional nodal analysis include: • Results are offered only at a snapshot, not as a function of time. • Inflow-performance-relationship (IPR) models are limited, with none available for shale gas wells. • Analysis is performed on a well-by-well basis, with no account of multiwell interference. We propose a new nodal-analysis method that enables the study of transient production systems, such as unconventional reservoirs, with IPR models generated from a high-speed semianalytical reservoir simulator and outflow curves generated from a steady-state pipeline simulator. The use of analytical reservoir simulation allows accurate, reliable modeling of the real inflow system. The new approach studies the time-lapse behavior of the system, with consideration of production history and neighboringwell interference. This new method enables the study of transient deliverability at the wellhead, where the measurement is usually available, and shows the time-lapse relationship between wellhead pressure and production rate. We provide examples of wellhead deliverability and choke management and explain advantages of the method with case studies involving tight and shale wells. The method is also applied to design and optimize artificial lift in unconventional wells and to study the method’s validity over time. In addition, we discuss an example of operational well dynamics with timelapse nodal analysis. Furthermore, this new method generates discussion about some concepts that are often taken for granted—What should be the definition of IPR in a transient production system? On the IPR curve, is the zero-rate-pressure the reservoir pressure? Can IPR curves at two different timesteps cross each other? Finding the answers to these questions will help us better understand production systems. The commonly used productivity-index (PI) method is reviewed and compared with the new method. Results show that one should not use the PI method when well operational conditions change. Introduction Nodal analysis has long been a key method used to evaluate the performance of an integrated production system (Mach et al. 1979). Components in the system can include reservoir, completion, tubing string, subsurface safety valves, surface choke, flowline, and separator. Nodes are placed to segment the system; each segment is defined by different equations or correlations. One can select the solution node, for example, at the bottomhole location. Pressure drops or gains from the starting point are added until the solution node is reached, which gives the inflow capacity or inflow-performance relationship (IPR). The same calculation applies from the solution node to the endpoint, which gives the outflow capacity or tubing-performance curve. The intersection of the two curves gives the solution operating point. Traditional nodal analysis is “static” and considers a snapshot of the whole production life. It assumes a pseudosteady-state or steady-state condition of the system. The IPR correlations used to describe the inflow system are based on certain assumptions, and are available for limited models: for example, vertical well, horizontal well, and fractured well. The analysis is performed on a well-by-well basis and without considering the interference effect of a neighboring well. Transient Inflow-Performance Relationship (IPR) in Literature and Current Practice When it comes to production in tight or shale formulations, it is widely accepted that pseudosteady-state or steady-state IPR curves are not applicable, and a transient time-changing IPR is required. Meng et al. (1982) developed a transient IPR for vertically fractured wells that is based on a semianalytical model at constant-rate production, described by the following equation: m½ pwf ðtÞ ¼ m½pi ðtÞ mwD ðtDxf ; Fcd Þ 1424 qg ðtÞ T ; kh ð1Þ where mðpÞ is real-gas pseudopressure, pwf is flowing bottomhole pressure, pi is initial pressure, mwD is the dimensionless pseudopressure-drop solution of a fractured vertical well, tDxf is dimensionless time, Fcd is dimensionless conductivity, qg is gas flow rate, and T is reservoir temperature. With Eq. 1, one can compute the IPR at any timestep t when given the input parameters. These IPRs and the outflow curve at different wellhead pressures are shown in Fig. 1. With this transient IPR, two things are noteworthy: • pwf ¼ pi if qg ¼ 0, meaning that IPR always starts from ð0; pi Þ. • IPR at time t is predetermined and not affected by the production history before time t. We will compare those findings with the results from the new approach proposed by this paper in later sections. Transient IPR curves for other models are found in the literature and in industry software. In general, if a dimensionless pressure solution is pD ðtÞ, then IPR at any time t is given by pwf ðtÞ ¼ pi 141:2qlB pD ðtÞ: . . . . . . . . . . . . . . . . . . ð2Þ kh Or denote the unit rate-pressure solution as pu ðtÞ ¼ 141:2lB pD ðtÞ; kh then, pwf ðtÞ ¼ pi qpu ðtÞ: . . . . . . . . . . . . . . . . . . . . . . . . . ð3Þ C 2016 Society of Petroleum Engineers Copyright V This paper (SPE 171768) was accepted for presentation at the Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 10–13 November 2014, and revised for publication. Original manuscript received for review 9 October 2014. Revised manuscript received for review 16 March 2015. Paper peer approved 28 April 2015. When production history is available, the current common practice to obtain the transient IPR is to calculate the well productivity index (PI) with the traditional equation February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 245 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Pressure (psi) 2,000 1,500 Pwh = 1,000 psi 1,000 Pwh = 800 psi Pwh = 600 psi 500 Pwh = 330 psi G F t7 t6 0 0 100 200 E D t5 t4 300 C B t3 400 A t2 500 600 t1 700 800 Fig. 1—Transient IPR (Meng et al. 1982). qðtÞ . . . . . . . . . . . . . . . . . . . . . . . . . ð4Þ pi pwf ðtÞ to capture the time-changing behavior of well productivity. Future IPRs are then obtained by simple extrapolation. As an example, a shale-oil production history is as shown in Fig. 2. The PI is then fitted with the red line and extrapolated for a future transient PI, as in Fig. 3. Another approach is to match the production history by reservoir simulation or rate-transient analysis (RTA), and to predict future production with the history-matched model and calculate a future PI with Eq. 4. Whether obtained by simple extrapolation, RTA, or simulation, the predicted IPRs are then used, for example, for future design of well completion and artificial lift. Note that, when discussing IPR, all methods described in this section use the same traditional PI concept (i.e., Eq. 4) involving initial reservoir pressure pi . For the sake of discussion, we name this group of methods the PI Method, which we will compare with the new method proposed in this paper. 2000 1800 pwf ¼ gðnÞ ðqÞ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð6Þ where gðnÞ represents the inflow curve at the nth timestep. The semianalytical reservoir model can be single well or multiwell. For a multiwell model, all neighboring well production is taken into account and has an impact on the IPR. The intersection of the inflow curve and outflow curve gives the solution of rate and BHP ðnÞ at the current timestep pwf and rate qðnÞ , which concludes the computation of this step: ( ( ðnÞ pwf pwf ¼ hðnÞ ðqÞ : . . . . . . . . . . . . . . . . . . . . ð7Þ ! ðnÞ pwf ¼ g ðqÞ qðnÞ Simulation then moves on to the next timestep. The whole process continues until it reaches the final timestep. Time-lapse nodal analysis gives the solution at requested timesteps, which, when taken all together, show the evolution of Qo 9000 1600 Oil-Production Cumulative (STB) 5000 6000 7000 8000 BHP 3000 4000 1400 1200 1000 800 Bottomhole Pressure (psi) 600 qo 200 1000 0 0 2000 Oil-Production Rate (STB/D) where hðnÞ represents the outflow curve at the nth timestep. Inflow performance, rather than with correlations as in traditional nodal analysis, is obtained by running a semianalytical reservoir simulation (Busswell et al. 2006; Gilchrist et al. 2007; Thambynayagam 2011; Zhou et al. 2013). It has the flexibility of modeling the transient behavior of real reservoir and well configurations. For instance, for a system (such as shown on the left in Fig. 5) with 2 years of production history, the objective of inflow simulation is to get the relationship between the BHP and rate for the current timestep. It is obtained by running simulation from the start of production, with all the historical rates, rates solved from previous timesteps, and an assumed rate for the current timestep. One can try a multitude of rates with a corresponding group of BHP responses. These rates and their BHP responses, represented on a plot of rate vs. BHP, define the inflow-performance relationship (IPR) for the current timestep, as on the right in Fig. 5. In a mathematical form, this is 10000 11000 12000 Time-Lapse Nodal-Analysis Principles In numerical reservoir simulation, one can define outflow curves for wells and run simulation with wellhead-pressure (WHP) control. Time-lapse nodal analysis mimics the numerical simulation approach and conducts nodal analysis through timestepping, as shown in Fig. 4. 400 Total Pages: 11 pwf ¼ hðnÞ ðqÞ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð5Þ Production Rate (Mscf/D) PIðtÞ ¼ Page: 246 Each timestep is composed of two primary parts—inflow simulation and outflow simulation. The intersection of the inflow curve with the outflow curve gives the operating point, which is the solution of the rate and bottomhole pressure (BHP), when given a wellhead pressure. The calculation then proceeds to the next timestep. Outflow simulation, as usual, is performed with a steady-state pipeline simulator with flow correlations. In this paper, outflow curves are simulated with Hagedorn and Brown (1965) correlation for vertical flow, Beggs and Brill (1973) revised correlation for horizontal flow, and Moody (1947) for single-phase flow. For a given WHP, it gives a relationship between production rate q and BHP pwf . This is expressed in a mathematical form, t 1 = 20 days t 2 = 35 days t 3 = 50 days t 4 = 100 days t 5 = 150 days t 6 = 250 days t 7 = 300 days 2,500 Stage: Oct 2011 Jan 2012 Jul 2012 Aug 2012 Oct 2012 Date Bottomhole Pressure Oil-Production Cumulative Oil-Production Rate Fig. 2—Production history of a shale oil well. 246 February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 247 Total Pages: 11 Pl = q/(Pi-Pwf ), (STB/D/psi) 100 Start 10 1 1 Next timestep 0.1 Multiwell inflow simulation 0.01 0.001 1 10 100 1,000 10,000 Outflow simulation 2 3 Time (days) Calculate operating point(s) Fig. 3—PI vs. time. production. For example, in Fig. 6, early-time and late-time IPR curves obtained from the simulator, together with the outflow curve throughout the time period, yield the production rate and BHP as functions of time, as shown in Fig. 7. Another important observation is that average reservoir pressure pavg , which is a required input for traditional nodal analysis, is not required in time-lapse nodal analysis. Instead, pavg is an output. Computational cost of time-lapse nodal analysis, thanks to the use of either analytical reservoir simulation or superposition, is minimal. Examples shown in this paper take no more than a fraction of a second. This is the case for the PI method too. Understanding Inflow-Performance Relationship (IPR) From Time-Lapse Nodal Analysis: Superposition Rule An IPR derived from semianalytical reservoir simulation shows distinct behavior, sometimes contradicting conventional nodalanalysis wisdom. To explain this behavior, we need to understand how semianalytical reservoir simulation is performed, and most importantly, to understand the superposition rule. The analytical solution of the pressure-diffusion equation follows this superposition rule: One can obtain the variable-rate solution from the constant-rate solution. Denote a unit-rate solution, or consider the pressure drop because of a unit-rate production Dp pi pwf ðtÞ ¼ pu ðtÞ. Then, the pressure of the varying rate qðtÞ is ðt DpðtÞ ¼ qðsÞ dpu ðt sÞ ds; . . . . . . . . . . . . . . . . . . . ð8Þ dt 0 or for discrete rate changes, Pressure (psia) Solved No 4 Last timestep Yes End Fig. 4—Time-lapse nodal-analysis work flow. pwf ðtÞ ¼ pi ½q1 pu ðtÞ þ N X ðqj qj1 Þpu ðt tj1 Þ: j¼2 ð9Þ Compare Eq. 9 with Eq. 3; we see that the productivity-index (PI) method takes no account of superposition. If rate history is known, qi ¼ qðti Þ; i ¼ 1 N 1, the pressure response at the Nth timestep is ( ) N1 X qj pu ðtN tj1 Þ pu ðtN tj Þ pwf ðtN Þ ¼ pi j¼1 qN pu ðtN tN1 Þ; ð10Þ which is essentially the IPR at the Nth timestep, with pwf ðtN Þ as a function of qN . From this equation, nXN1 we can see that this IPR curve starts o at a pressure p0 ¼ pi q ½p ðt tj1 Þ pu ðtN tj Þ , and j¼1 j u N the slope is pu ðtN tN1 Þ, as shown in Fig. 8. Besides the rate superposition just described, there is also pressure superposition: One can obtain the variable-bottomhole- New timestep BHP 2000 1000 WHP 400 300 200 Liquid Rate (STB/D) Gas Rate (Mscf/D) BHP Rate 500 0 2009 2010 2011 2012 Rate Date/Time Fig. 5—Calculated inflow (left), inflow curve, and outflow curve at timestep n (right). February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 247 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 248 Total Pages: 11 10,000 5,000 Pressure (psia) pwf 5,000 WHP 10 5 0 Liquid Rate (B/D) Pwf (psia) 10,000 pavg 10,000 Liquid Volume (MMSTB) Inflow curve Operating points Outflow curve 02/29/2012 14:02:30 03/01/2012 14:02:30 03/02/2012 14:02:30 03/03/2012 14:02:30 03/04/2012 14:02:30 03/05/2012 14:02:30 03/06/2012 14:02:30 03/13/2012 14:02:30 03/29/2012 14:02:30 04/28/2012 14:02:30 05/28/2012 14:02:30 06/27/2012 14:02:30 07/27/2012 14:02:30 08/26/2012 14:02:30 09/25/2012 14:02:30 10/25/2012 14:02:30 11/24/2012 14:02:30 12/24/2012 14:02:30 01/23/2012 14:02:30 02/22/2012 14:02:30 qo 30,000 Qo 25,000 Apr 2012 0 Jul Oct Jan 2013 Date/Time Fig. 7—Time-lapse nodal-analysis results. 0 2e5 4e5 6e5 Liquid Rate (B/D) BQN is the average reservoir pressure, /hct A 70:6Bl 4A ln c and the slope is the well PI, according to Eq. kh e CA rw2 A-25 in Blasingame and Lee (1986). Therefore, for such systems, time-lapse nodal analysis reduces to traditional nodal analysis. where p0 ¼ pi 0:2339 Fig. 6—Inflow and outflow curves in time-lapse nodal analysis. pressure (BHP) solution from the constant-BHP solution by superposition, with the following equation: ðt dqu ðt sÞ ds; . . . . . . . . . . . . . . ð11Þ qðtÞ ¼ pi pðsÞ dt 0 or for discrete BHP changes, ( ) N1 X qðtN Þ ¼ ðpi pj Þ qu ðtN tj1 Þ qu ðtN tj Þ j¼1 ðpi pN Þqu ðtN tN1 Þ: . . . . . . . . . . . . . . ð12Þ The effect of gas nonlinearities is handled by gas pseudopressure and gas pseudotime. For the sake of simplicity, this is not discussed in this paper. The superposition rule holds true for tight formations and conventional reservoirs with high-to-medium permeability. It holds for a transient, pseudosteady-state, or steady-state system. For a well centered in a bounded, circular conventional reservoir, Eq. 10 becomes the following [according to Eq. A-42 in Blasingame and Lee (1986)]: BQN 70:6Bl 4A ln c ; qN pwf ðtN Þ ¼ pi 0:2339 kh e CA rw2 /hct A ð13Þ Discussion of Inflow-Performance-Relationship (IPR) Curve The new concept of time-lapse nodal analysis brings new ideas to consider in how we think of the inflow-performance curve. Some examples follow. Pressure at Zero Rate Is the Buildup Pressure, Not Necessarily the Reservoir Pressure. According to the conventional IPR concept, pressure at zero rate is the average reservoir pressure. But from the preceding analysis, we can see this pressure is actually the buildup pressure for a shut-in (rate equals zero). In high-permeability conventional reservoirs, it may build up to the actual reservoir pressure; then, the meaning of pressure at zero rate converges to the meaning in the conventional concept. For unconventional shale reservoirs, the pressure may not be able to reach reservoir pressure within the duration of the shut-in. One Can Define Productivity Index (PI) With Buildup Pressure. According to Eq. 10 and as shown in Fig. 8, the slope of the curve is pu ðtN tN1 Þ. If the timestep size, Dt ¼ tN tN1 , is constant, the slope will be identical at all times, making the well PI constant. For a conventional reservoir system, as described in Eq. 13, this corresponds to a constant-productivity well with reducing reservoir pressure. For shale wells, if we propose a new way of defining the well PI, PI ¼ pi – {Σ N–1 j =1 qj [pu } (tN – tj–1) – pu (tN – tj )] pu (tN – tN–1) pwf 0 q Fig. 8—IPR from superposition. 248 qðtÞ ; . . . . . . . . . . . . . . . . . . . . . . . ð14Þ pbu ðtÞ pwf ðtÞ where pbu is the buildup pressure at q ¼ 0; this corresponds to a constant-productivity well with reducing buildup pressure. This new definition, compared with the traditional definition with respect to initial reservoir pressure or average reservoir pressure, is more practical and suitable to model shale wells, because buildup pressure is measurable. IPR Depends On Production History. The IPR of a well is generally considered a predefined attribute independent of outflow curves and production history. For example, in Fig. 9, a set of IPRs is predefined at three times and used for analysis for any outflow curves (e.g., tubing-performance curves TPC1 and TPC2), which then gives the two production predictions, as shown on the right. From time-lapse nodal analysis, however, IPR at one timestep is the result of superposition and is related to production history before this timestep, which, in turn, is February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 249 Total Pages: 11 TPC1 q TPC2 pwf IPR@t3 0 IPR@t2 q@TPC2 IPR@t1 q@TPC1 q t Fig. 9—IPR curves independent of outflow curves. 12,000 12,000 Op. points Op. points 10,000 10,000 TPC–1 TPC–2 IPR, t = 30 days IPR, t = 30 days 8,000 IPR, t = 60 days IPR, t = 90 days 6,000 IPR, t = 120 days IPR, t = 150 days 4,000 IPR, t = 180 days BHP (psi) BHP (psi) 8,000 IPR, t = 60 days IPR, t = 90 days 6,000 IPR, t = 120 days IPR, t = 150 days IPR, t = 180 days 4,000 IPR, t = 210 days IPR, t = 210 days 2,000 IPR, t = 240 days IPR, t = 240 days 2,000 IPR, t = 270 days IPR, t = 270 days 0 IPR, t = 300 days 0 500 1,000 1,500 Rate (B/D) IPR, t = 300 days 0 2,000 2,500 0 500 1,000 1,500 Rate (B/D) 2,000 2,500 Fig. 10—IPR curves dependent on outflow curves: IPR curves for TPC1 (left) and TPC2 (right). dependent on the outflow curve. As a result, IPR curves cannot be predefined and can only be determined during the analysis. For example, in Fig. 10, IPR curves for TPC1 (high, constant pwf ) are higher than those for TPC2 (low, constant pwf ). In practical application, it means that • For new well, IPR calculation for timestep n needs to take into account the superposition effect of the production rate calculated from timestep 1 to n 1. • For wells with long history, similar to the technique used in well-test analysis, the rate history occurred a long time before the analyzed period could be simplified and the rate variations that happened immediately before should be taken more accurately (Bourdet 2002). 4,500 4,500 4,000 4,000 3,500 3,500 Op. points 3,000 TPC 2,500 IPR, t = 10 days IPR, t = 20 days 2,000 IPR, t = 30 days IPR, t = 40 days (dt = 10 days) IPR, t = 50 days 1,500 1,000 IPR, t = 60 days 500 BHP (psi) BHP (psi) IPR Is a Function of Timestep. Well productivity is normally considered a well property associated with time (e.g., 05/26/2014) but not with timesteps (e.g., delta time, 1 day, 1 month, and 1 year). For example, one always asks this kind of question: What is my well productivity today? From time-lapse nodal analysis, however, we can see that IPR is strongly related to the timesteps used. With a constant unit rate, pressure drop always increases as time increases [i.e., pu ðDt1 Þ < pu ðDt2 Þ if Dt1 < Dt2 ]; therefore, we will have low productivity at large timesteps, regardless of the time itself. Results shown on the left plot in Fig. 11 are from the nodal analysis with a constant pwf ¼ 1; 500 psi at equal timesteps, t ¼ 10; 20; 30; 40; 50; 60 days. Because the timestep remains the same (10 days), all IPRs share the same slope. The right plot in Fig. 11 shows results of the same nodal analysis, but one timestep (from the time interval 30 days to 40 days) is divided into five substeps of 2 days each, t ¼ 32; 34; 36; 38; 40 days. The IPRs at these five substeps show a distinctly different slope. Both analyses give the same result, as shown on the left plot of Fig. 12. However, as shown on the right in Fig. 12, the IPR curve at t ¼ 40 days with the 10-day step is steeper, or appears as lower Op. points TPC IPR, t = 10 days 3,000 IPR, t = 20 days 2,500 IPR, t = 30 days 2,000 IPR, t = 32 days 1,500 IPR, t = 36 days IPR, t = 34 days IPR, t = 38 days IPR, t = 40 days (dt = 2 days) IPR, t = 50 days 1,000 500 IPR, t = 60 days 0 0 1,000 2,000 3,000 Rate (B/D) 4,000 0 0 1,000 2,000 3,000 4,000 5,000 Rate (B/D) Fig. 11—Results of time-lapse nodal analysis with timestep of 10 days (left), with substeps of 2 days between 30 and 40 days (right). February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 249 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 250 Total Pages: 11 4,500 2,500 4,000 3,500 3,000 1,500 BHP (psi) Rate (B/D) 2,000 1,000 2,500 Op. points TPC IPR, t = 10 days (dt = 10 days) IPR, t = 10 days (dt = 2 days) 2,000 1,500 500 1,000 0 0 10 20 30 40 Time (days) Rate(dt = 10 days) 50 60 500 70 0 0 Rate(with substeps) 1,000 2,000 3,000 Rate (B/D) 4,000 Fig. 12—Production rate (left), and IPR comparison (right). 100 10 1 0 200 400 600 800 1,000 1,200 1,800 1,600 1,400 1,200 1,000 800 600 400 200 0 1,200 1,000 800 600 400 200 0 BHP (psi) 1,000 Rate (B/D) Rate (STB/D) 10,000 TLNA-Rate PI-Rate TLNA-Pwf PI-Pwf 0 50 100 150 200 250 300 350 Time (days) Time (days) Fig. 14—PI method and time-lapse nodal analysis give the same result. Fig. 13—Rate profile at constant BHP. productivity, than that with the 2-day step, and the two curves intersect at the operating point. The dependence of the IPR on timestep is important because it means we cannot use the same set of IPRs for long-term production prediction with a monthly timestep and for operational well dynamics when a timestep is on the order of minutes and hours. This is elaborated in a later section. Comparison Between Time-Lapse Nodal-Analysis and the Productivity-Index (PI) Methods We use a synthetic example under a variety of scenarios to compare the PI-analysis method with the time-lapse nodal-analysis method. The synthetic model is a shale oil well in a reservoir with initial reservoir pressure of 4,200 psi. The well-production profile at constant pwf ¼ 2,000 psi is as shown in Fig. 13. For the PI method, one can simply calculate the transient PI at each month by Eq. 4: PI ¼ qðtÞ=ð4; 200 2; 000Þ. However, in the time-lapse 1,400 2,500 2,000 1,000 800 1,500 600 1,000 400 500 200 0 0 BHP (psi) Rate (B/D) 1,200 TLNA-Rate PI-Rate TLNA-Pwf PI-Pwf 0 50 100 150 200 250 300 350 Time (days) Fig. 15—PI method and time-lapse nodal analysis give similar result. 250 nodal-analysis method, PI at timestep N is calculated by pressuresuperposition (Eq. 12), considering the history pj ; j ¼ 1 N 1. Flat Tubing-Performance Curve. Under constant bottomhole pressure (BHP), the two methods give the same result (Fig. 14). Upward Tubing-Performance Curve. Even for a slightly upward-trending tubing-performance curve, the two methods give similar results (Fig. 15). The PI method gives the same or similar results as time-lapse nodal analysis for two previous cases, because the condition under which the system is evaluated, constant BHP [flat tubing-performance curve (TPC)] and slightly upward-trending TPC, is within the assumption of the PI method, that is, stable production condition. When production condition changes, however, as in the next two cases, the PI method fails to capture the right behavior. Effect of Shut-in. Suppose the well produced for 3 months and then was shut down for 3 months and opened again afterward under the same pwf before the shut-in. Production after the shut-in would increase because of the charging of the formation during shut-in. The time-lapse nodal analysis correctly captures this behavior, compared with the result from numerical reservoir simulation in red squares (Fig. 16). In contrast, the PI method fails to account for the shut-in, giving exactly the same production trend despite the shut-in. Also, examine pwf during the shut-in, from 90 days to 180 days; the PI method moves pwf back to the initial pressure of 4,200 psi, but the time-lapse analysis correctly models the pressure buildup. Two Tubing-Performance Curves. When the operational parameters of the well change—for example, when the choke is opened wider or artificial lift is installed and the BHP is lowered February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 1,800 1,600 1,400 1,200 1,000 800 600 400 200 0 0 50 100 150 Time (days) Stage: 200 Page: 251 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 0 250 Total Pages: 11 TLNA-Rate BHP (psi) Rate (B/D) J171768 DOI: 10.2118/171768-PA Date: 3-February-16 PI-Rate Rate from Res. Sim. TLNA-Pwf PI-Pwf Fig. 16—PI method fails to account for the shut-in. 1,400 2,500 2,000 1,000 800 1,500 600 1,000 400 500 200 0 BHP (psi) Rate (B/D) 1,200 0 TLNA-Rate PI-Rate TLNA-Pwf PI-Pwf 0 50 100 150 200 250 300 350 Time (days) Fig. 17—PI method fails to capture the rate increase caused by lower BHP. from 2,000 psi to 1,100 psi, we see from Fig. 17 that the timelapse nodal-analysis method captures the rate increase correctly, and the PI method underestimates the increase. The PI method yields a good approximation when well operational conditions remain unchanged. If there are changes in, for example, choke positions, lifting mechanisms, or shut-ins, the PI method will greatly underestimate the production after the change, and one should not use that method. Application of Time-Lapse Nodal Analysis The following examples show the application of time-lapse nodal analysis. Although the examples are either tight or shale wells, the principles apply to studies of any type of wells when the traditional nodal analysis falls short of expectations. Wellhead Deliverability. Well deliverability is usually referenced at bottomhole conditions in terms of the productivity index (PI) (STB/D/psi). However, bottomhole deliverability is difficult to use operationally if downhole-pressure measurement is not available in relevant time. Because wellhead pressure (WHP) is readily available most of the time, wellhead deliverability will shed light on well performance and guide well operations. For example, Fig. 18 is the crossplot of wellhead pressure vs. gas rate for a tight gas well. The color scale represents time—blue refers to time back in history and red refers to the most-recent time. The dots are aligned in a set of straight lines corresponding to different choke positions. What is the wellhead deliverability today? If it is Curve A, the well has much more production potential than if it is Curve B. To predict wellhead-deliverability curves, we can history match the production with rate-transient analysis and predict into the future at different wellhead pressures with time-lapse nodal analysis, as shown in Fig. 19. Keeping the same WHP of 1,689 psi would continue the rate-declining trend (brown curve in the NA sensitivity plot with History Curve 1 Wellhead Pressure (psia) 4,000 2,000 Wellhead Pressure Time A 0 03/08/2009 B 1,000 2,000 Production Rate_1;qg (Mscf/D) 03/22/2010 3,000 Fig. 18—WHP vs. rate, and wellhead-deliverability curve. February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 251 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 252 Calc_BHP_Go Calculated reservoir pressure Wellhead pressure Production Rate_1:qg Press (Line style) Pwf (Line style) qg (Line style) Wellhead pressure (Line style) Wellhead pressure: 1,000 psia (Color) Wellhead pressure: 1,250 psia (Color) Wellhead pressure: 1,689 psia (Color) Wellhead pressure: 250 psia (Color) Wellhead pressure: 750 psia (Color) 4,000 Pressure (psia) Total Pages: 11 2,000 Gas Rate (Mscf/D) 4,000 2,000 Sep Oct Nov Dec Jan 2010 2009 Feb Mar Apr May Jun Date/Time Fig. 19—Production history matching (before February 2010) and prediction with time-lapse nodal analysis with different WHPs. bottom pane of the figure); lowering WHP to 250 psi would boost the rate from 1 MMscf/D to 4 MMscf/D before it falls back again because of depletion. One can plot together WHP vs. predicted rate at different timesteps, as in Fig. 20, which shows the wellhead deliverability. One can use these curves in well operations— for example, in choke management. Shale-Oil Well-Production Performance. A 4,600-ft-long shale oil well at a depth of 6,500 ft has 28 hydraulic fractures and a tubing size of 2.875 in. For simplicity, fractures are equidistant and of equal length. We analyzed the sensitivity of WHP with the time-lapse nodal-analysis method. Fig. 21 shows the inflow-performance relationship (IPR) and tubing-performance curves (TPCs) at eight timesteps for two WHPs: 100 and 500 psi. The upper TPC is for WHP ¼ 500 psi; the lower one is for WHP ¼ 100 psi. The IPR at the first timestep (the outermost one) is the same for both pressures. Starting from the second timestep (because the history from previous timesteps is different), the IPR curves differ for these two cases. In Fig. 22, the result, plotted as a function of time, shows that cumulative production at low WHP is approximately twice that of the cumulative production at high WHP. Production at WHP ¼ 500 psi would reach the economic limit of 20 B/D in 7 months; for WHP ¼ 100 psi, the limit would be reached NA sensitivity plot with History Curve 1 Wellhead pressure Wellhead Pressure (psia) 4,000 Operation Points (Line style) Wellhead pressure: 1,000 psia (Color) Wellhead pressure: 1,250 psia (Color) Wellhead pressure: 1,689 psia (Color) Wellhead pressure: 250 psia (Color) Wellhead pressure: 750 psia (Color) * After the end of the production history (January 30, 2010) 2,000 11 days* Pavg = 2,190 psia 0 91 days Pavg = 1,443 psia 123 days Pavg = 1,276 psia 61 days Pavg = 1,617 psia 30 days Pavg = 1,940 psia 08/21/2009 2,000 Production Rate_1;qg (Mscf/D) Time 06/01/2010 4,000 Fig. 20—Wellhead-deliverability curves. 252 February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 253 Total Pages: 11 Inflow curve Pressure (psia) Inflow Wellhead Pressure: 100 psia Inflow Wellhead Pressure: 500 psia Operating points Outflow Wellhead Pressure: 100 psia Outflow Wellhead Pressure: 500 psia 3,000 12/04/2013 00:00:00 12/24/2013 00:00:00 07/23/2014 00:00:00 03/24/2014 00:00:00 3,000 Pres Pwf qo Qo Wellhead Pressure Wellhead Pressure: 100 psia Wellhead Pressure: 500 psia 2,000 1,000 07/22/2014 00:00:00 03/19/2015 00:00:00 07/11/2016 00:00:00 0 Liquid Volume (MMSTB) 2,000 1,000 0.04 Qo@pwh = 100 psi Liquid Rate (B/D) Pwf (psia) 02/26/2019 00:00:00 102 0.02 Qo@pwh = 500 psi qlim : 20 B/D 101 0.00 2014 2015 2016 2017 Date/Time 2018 2019 Fig. 22—Time-lapse nodal-analysis results. 0 200 400 Liquid Rate (B/D) 600 Fig. 21—Sensitivity analysis of wellhead pressure for a shale oil well. in 2 years. The bottomhole pressure (BHP) of the scenario with WHP ¼ 500 psi is approximately 3,000 psi, and it is approximately 2,500 psi when the WHP ¼ 100 psi. Artificial-Lift Design and Optimization for Shale Well. Next, we will apply time-lapse nodal analysis and model the production of a shale oil well under artificial lift and optimize the performance of the pump. The shale oil well is the one used in Fig. 13, and the figure shows the well production under constant pwf ¼ 2,000 psi. The well has been under natural flow for 100 days, and we want to investigate artificial-lift methods for the future. If the same natural flow is maintained, production follows the same trend, and both methods give the same result (Fig. 23). If an electrical submersible pump (ESP) is installed, in this case a REDA DN1750 with 322 stages, and we set the frequency at 60 Hz, the production jumps to 1,200 B/D and falls back to 800 B/D in 100 days, as in Fig. 24. On average, the PI method underestimates the rate by 200 B/D. One should optimize the production rate through a pump so that it falls within the recommended operating range to avoid downthrust and upthrust. Variable-speed drives (VSDs) can vary the speed of pumps and enable pumps to operate across a wider range, improve performance, and optimize productivity. So, next, we investigate the effect of different frequencies on the resulting production profile. As in Fig. 24, if the frequency is set at 45 Hz, the production jumps to 1,100 B/D and falls back to 600 B/D in 100 days. On average, the PI method underestimates the rate by 160 B/D. If the frequency is set at 30 Hz, the rate jumps to 800 B/D and falls back to 500 B/D in 100 days. On average, the PI method underestimates the rate by 85 B/D. In general, the effect of an ESP on production Rate (B/D) 500 201 1,200 1,000 800 600 400 0 1,000 101 151 Time (days) 1,400 200 1,500 51 1,600 2,500 2,000 1 Operational Well Dynamics. The preceding examples show the application of time-lapse nodal analysis on long-term production behavior, on the order of months to years. The method also has application in short-term operational well dynamics, those with a time scale of hours to days. Timesteps can be as small as seconds. As previously discussed, the timestep is a key parameter that determines the slope of the IPR curves, and the use of a wrong IPR would give wrong results. Full study of well dynamics would require the link of transient inflow simulation with transient outflow simulation. In this paper, we explain only the importance of having the right transient inflow model by linking with a steadystate outflow simulation. A full study is the topic of another paper. As an example, for the shale well in Fig. 23, under natural flow after 100 days, the choke is opened at 8/64, 16/64, 32/64, and 64/ 64 in. sequentially, each for 1 hour in a 4-hour cycle. The well behavior is modeled with both methods, as shown in Fig. 26. Notice the IPRs from time-lapse nodal analysis and those from the PI method: Well productivity from time-lapse analysis is 12.5 B/D/ 3,000 BHP (psi) 1,600 1,400 1,200 1,000 800 600 400 200 0 and the impact of a different frequency are underestimated by the PI method. The production rate at different frequency values, modeled with both methods, is overlayed with the pump VSD curves in Fig. 25. The rate predicted by the PI method is totally outside the recommended operating range, whereas the rate profile from time-lapse nodal analysis falls into the efficient range for 45 and 30 Hz. This means that the use of the PI method would result in a false optimization of the pump. Rate (STB/D) 0 TLNA-Rate PI-Rate Rate history TLNA-Pwf PI-Pwf Pwf history 0 251 Fig. 23—Well follows the same trend with natural flow. 1 50 100 150 Time (days) 200 Hist NF TLNA-30 Hz PI-60 Hz TLNA-60 Hz TLNA-45 Hz TLNA-45 Hz PI-30 Hz 250 Fig. 24—Production profile at 60 Hz (red), 45 Hz (green), and 30 Hz (blue), from time-lapse nodal analysis (solid line) and PI method (dashed line). February 2016 SPE Journal ID: jaganm Time: 22:52 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 253 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Stage: Page: 254 Reda - DN1750 Stages = 322 Reda - DN1750 Stages = 322 @90 Hz @85 Hz @80 Hz @75 Hz Head (ft) Head (ft) @90 Hz 18,000 16,000 14,000 12,000 10,000 @70 Hz @65 Hz @60 Hz 8,000 6,000 @55 Hz @50 Hz @45 Hz @40 Hz @35 Hz @30 Hz 4,000 2,000 500 Total Pages: 11 18,000 16,000 14,000 12,000 10,000 8,000 6,000 @85 Hz @80 Hz @75 Hz @70 Hz @65 Hz @60 Hz @55 Hz @50 Hz @45 Hz @40 Hz @35 Hz @30 Hz 4,000 2,000 1,000 1,500 2,000 2,500 Flow Rate (B/D) 3,000 3,500 500 1,000 1,500 2,000 2,500 Flow Rate (B/D) 3,000 3,500 Fig. 25—Pump VSD curves overlayed with the production profile at different frequency, from PI method (left) and time-lapse nodal analysis (right). 5,000 8/64 32/64 16/64 4,500 4,000 3,500 64/64 The huge difference in the IPRs and the resulting pressure and rate profile have a big impact in multiphase-flow regimes and dynamic well behavior. This topic will be further investigated with both the transient inflow model and transient outflow model in a separate paper. BHP (psi) 3,000 2,500 2,000 IPR from timelapse nodal analysis 1,500 1,000 IPR from PI method 500 0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000 Rate (B/D) Fig. 26—IPR difference between time-lapse nodal analysis and PI method. psi and only 0.285 B/D/psi according to the PI method, a difference of 44 times. For the resulting rate and pressure profile, as shown in Fig. 27, the PI method predicts a BHP fluctuation of approximately 1,800 psi, but the time-lapse method predicts that the BHP fluctuation will be less than 200 psi. The rate from the PI method fluctuates in a range of 500 B/D, whereas the time-lapse method predicts a fluctuation range of 1,500 B/D in this 4-hour cycle. 2,000 1,800 Conclusions The time-lapse nodal-analysis advantageous characteristics are • Has application in both unconventional and conventional resources. • Captures the transient behavior of the reservoir. • Models inflow performance relationship (IPR) with semianalytical reservoir simulation or superposition; no IPR correlations are required. • Evaluates production performance as a function of time, not as a static snapshot. • Captures wellhead deliverability as a function of time. • Is suitable for study of operational well dynamics and artificiallift design and optimization. In contrast, the traditional PI-method characteristics are • Can only be used when well-operational conditions remain unchanged. • Underestimates production, and one should not use it for artificial-lift design and optimization for wells in tight/shale formations. • Overestimates bottomhole-pressure fluctuation and underestimates rate fluctuation. 3,500 3,000 1,600 1,200 2,500 2,000 1,000 800 600 1,500 1,000 400 200 500 BHP (psi) Rate (B/D) 1,400 TLNA-Rate PI-Rate Rate TLNA-Pwf PI-Pwf Pwf 0 100.00 100.05 100.10 100.15 100.20 100.25 100.30 100.35 100.40 Time (days) Fig. 27—PI method overestimates pwf fluctuation and underestimates rate fluctuation. 254 February 2016 SPE Journal ID: jaganm Time: 22:53 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 J171768 DOI: 10.2118/171768-PA Date: 3-February-16 Nomenclature A ¼ reservoir area, ft2 B ¼ volume factor ct ¼ total compressbility, 1/psi CA ¼ Dietz shape factor Fcd ¼ fracture conductivity, dimensionless h ¼ pay-zone thickness, ft k ¼ formation permeability, md mwD ¼ pseudopressure drop of a fractured vertical well, dimensionless pavg ¼ average reservoir pressure, psi pbu ¼ buildup pressure, psi pi ¼ initial reservoir pressure, psi pu ¼ unit rate pressure, psi/(B/D) pwf ¼ flowing bottomhole pressure, psi pwh ¼ wellhead pressure, psi q ¼ production rate, B/D or Mscf/D qo ¼ oil-production rate, B/D qu ¼ unit drawdown rate, (B/D)/psi Q ¼ cumulative production Qo ¼ cumulative oil production, bbl rw ¼ wellbore radius, ft t ¼ time, days tDxf ¼ dimensionless time T ¼ temperature c ¼ 0.577216, Euler’s constant l ¼ viscosity, cp / ¼ porosity, fraction Acknowledgments The authors would like to thank Schlumberger for supporting this work and giving permission to publish this paper. References Beggs, H. D. and Brill, J. P. 1973. A Study of Two-Phase Flow in Inclined Pipes. J Pet Technol 25 (5): 607–617. SPE-4007-PA. http://dx.doi.org/ 10.2118/4007-PA. Blasingame, T. A. and Lee, W. J. 1986. The Variable-Rate Reservoir Limits Testing. Presented at the SPE Permian Basin Oil and Gas Recovery Conference, Midland, Texas, USA, 13–14 March. SPE-15028-MS. http://dx.doi.org/10.2118/15028-MS. Bourdet, D. 2002. Well Test Analysis: The Use of Advanced Interpretation Models. Elsevier. Busswell, G., Banerjee, R., Thambynayagam, R. M. K. et al. 2006. Generalized Analytical Solution for Reservoir Problems With Multiple Wells and Boundary Conditions. Presented at the Intelligent Energy Conference and Exhibition, Amsterdam, The Netherlands, 11–13 April. SPE-99288-MS. http://dx.doi.org/10.2118/99288-MS. Gilchrist, J. P., Busswell, G., Banerjee, R. et al. 2007. Semi-analytical Solution for Multiple Layer Reservoir Problems With Multiple Vertical, Horizontal, Deviated, and Fractured Wells. Presented at the International Petroleum Technology Conference, Dubai, UAE, 4–6 December. IPTC-11718-MS. http://dx.doi.org/10.2523/11718-MS. Hagedorn, A. R. and Brown, K. E. 1965. Experimental Study of Pressure Gradients Occurring During Continuous Two-Phase Flow in Small Diameter Vertical Conduits. J Pet Technol 17 (4): 475–484. SPE-940PA. http://dx.doi.org/10.2118/940-PA. Stage: Page: 255 Total Pages: 11 Mach, J., Proano, E., and Brown, K. E. 1979. A Nodal Approach for Applying Systems Analysis to the Flowing and Artificial Lift Oil or Gas Well. SPE-8025-MS. http://dx.doi.org/10.2118/8025-MS. Meng, H. Z., Proano, A. P., Buhidma, I. M. et al. 1982. Production Systems Analysis of Vertically Fractured Wells. Presented at the SPE/ DOE Unconventional Gas Recovery Symposium, Pittsburgh, Pennsylvania, USA, 16–18 May. SPE/DOE-10842. SPE of AIME. http:// dx.doi.org/10.2118/10842-MS. Moody, L. 1947. An Approximate Formula for Pipe Friction Factors. Trans. ASME 69: 1005. Thambynayagam, R. M. K. 2011. The Diffusion Handbook: Applied Solutions for Engineers. New York: McGraw-Hill Professional. Zhou, W., Samson, B., Krishnamurthy, S. et al. 2013. Analytical Reservoir Simulation and Its Applications to Conventional and Unconventional Resources. Presented at the EAGE Annual Conference and Exhibition Incorporating SPE EUROPEC, London, UK, 10–13 June. SPE164882-MS. http://dx.doi.org/10.2118/164882-MS. Conversion Factors bbl 1.589 873 cp 1.0* ft 3.048* in. 2.54* psi 6.894 757 E01 ¼ m3 E03 ¼ Pas E01 ¼ m Eþ00 ¼ cm Eþ00 ¼ kPa *Conversion factor is exact. Wentao Zhou is Production Product Champion with Schlumberger, based in Houston. In his current role, he works on software product development, product marketing, project consulting, and technical training. Zhou’s products cover areas such as analytical reservoir simulation, pressure- and rate-transient analysis, unconventional-resources production, artificial lift, reservoir modeling and simulation, and field-development planning. He joined Schlumberger in 2006. Zhou holds an MSc degree in petroleum engineering from Stanford University and an MSc degree in fluid mechanics from Peking University, China. Raj Banerjee is Advisor with Schlumberger. Based in Houston, he is currently responsible for the development and deployment of digital-oilfield-software solutions globally. Banerjee is well-published and has authored or coauthored more than a dozen patents. He holds a PhD degree in applied mathematics from Imperial College London. Banerjee also holds an MS degree in petroleum engineering from the University of Tulsa and a BS degree in petroleum engineering from Indian School of Mines. Eduardo A. Proano is Advisor, Reservoir Discipline Career Manager, for Schlumberger. He has been with the company for more than 27 years. Proano’s expertise includes production optimization, well intervention, online production surveillance and diagnosis, rapid field/asset wide evaluations for production enhancement, asset-management applications for well productivity, reservoir management, and field-development plan fast-track solutions. He holds BS and MS degrees in petroleum engineering and a Masters of Engineering Management degree from the University of Tulsa. February 2016 SPE Journal ID: jaganm Time: 22:53 I Path: S:/J###/Vol00000/150054/Comp/APPFile/SA-J###150054 255