Documente Academic
Documente Profesional
Documente Cultură
Submitted by: Mike Mosher 200816593 Nicholas House 200814127 Christopher Furlong 200931863 Evan Hipditch 200738284
Executive Summary
This document highlights the second phase of the design work that MUNRS has undertaken towards achieving their goal of testing a new method of enhanced oil recovery. This method, in which produced gas is separated into light and heavy components, for both gas lift and gas injection respectively, is being modeled through industry-standard simulation techniques. Once MUNRS had proven that the minimum miscibility pressure (MMP) of the produced gas and the reservoir fluids were not similar (i.e. the produced gas would not be effective for gas injection in its current state), the group made effective use of a two-phase oil and gas separator. By optimizing the temperature and pressure of the first stage of the separator, and effectively removing the right quantity of light components from the fluid, MUNRS was able to remove a composition of rich gas from the separators second stage that when tested, had an MMP that was very close to the pressure of the Norne reservoir the reservoir for this case study. Since these were roughly the same pressure (297.8 and 298 bar respectively), it was determined that a developed miscibility would be possible with this fluid. Once the injection composition was determined, MUNRS then modeled the Inflow Performance Relationship (IPS) and Vertical Lift Performance Curves (VLP) using production equipment standards found in modern oil and gas installations. This detailed analysis confirmed that gas lift will be of great benefit to the production from wells in the case study, and that the test case well was functioning properly.
The final step undertaken in this design phase by MUNRS was to model a base case block model of a homogeneous reservoir so that the pre-determined fluid compositions and lift analysis could be tested in the context of a real oil and gas reservoir. This model will show results, as they would be obtained from an oil and gas reservoir in industry. Moving forward, MUNRS has decided to focus on using the reservoir block model, with the refined inputs of fluid composition, and gas lift analysis to obtain final results for the improved performance of the case study. An economic analysis will then be completed, showing a large-scale breakdown of the costs involved with using a new technology such as this, in a modern oil and gas facility.
Table of Contents
1.0 INTRODUCTION 2.0 PROBLEM DEFINITION 3.0 DESIGN SPECIFICATION AND CONSTRAINTS 3.1 TIME 3.2 RESEARCH 3.3 SIMULATION AND MODELING 4.0 MINIMUM MISCIBILITY PRESSURE AND FLUID SEPARATION 4.1 MMP TEST 4.2 RECOMBINATION TEST 4.3 SEPARATOR TEST 5.0 SEPARATOR OPTIMIZATION 5.1 MODELING THE RELATIONSHIP P,T AND MMP 6.0 VERTICAL LIFT PERFORMANCE 7.0 RESERVOIR SIMULATION 8.0 PROJECT PROGRESS AND MANAGEMENT 81 PROJECT MANAGEMENT 8.2 CHANGES IN SCOPE 8.3 MOVING FORWARD 1 2 2 2 3 3 3 3 5 6 8 8 11 14 19 19 19 20
Table of Figures
Figure 1 - Response Surface of Intial MMP Optimization...................................................................... 9 Figure 2 - Response Surface for Refined MMP Optimization ............................................................ 10 Figure 3 - ProsperTM IPR Curve ..................................................................................................................... 12 Figure 4 - ProsperTM VLP Curces .................................................................................................................. 13 Figure 5 - Block Model ..................................................................................................................................... 18
List of Tables
Table 1 MMP Volumes for Intial Seperator ............................................................................................. 7 Table 2 Refined Table of MMPs from Refined Seperator Test ..................................................... 10 Table 3 Constant Properties....................................................................................................................... 15 Table 4 Reservoir Fluid Components ..................................................................................................... 15
Appendices
Appendix A Updated Gantt Chard Appendix B Selected Values from PVTsimTM Appendix C Table of values used in ProsperTM
1.0 Introduction
ENGI 8926 Mechanical Design Project II is the second of two capstone design courses in the Mechanical discipline. Building on skills developed in the first, student teams each choose a unique design challenge and then proceed to generate a solution. MUNRS (Memorial University Reservoir Solutions) is a group of four mechanical engineering students from Memorial University. This term, they have been tasked with the design goal of testing a new method of enhanced oil recovery. This new method will make use of produced gas for both gas injection and gas lift, in the most effective way that the group sees possible by splitting the valuable fluid into two, more effective compositions. This document will focus on the progress MUNRS has made towards accomplishing the prescribed goal. MUNRS is being co-supervised this term by Dr. Lesley James, and Dr. Thormod Johansen. Supervision will consist of weekly meetings, and communication through email.
3.1 Time
For this reporting phase of the project, time is a significant design constraint, with a full reporting and presenting stage due the week of March 7 . One time constraint is the availability of Teaching Assistants who will help us with the technical nature of the project. Lab availability will also be a time constraint, as we are sharing
th
3.2 Research
With any project, there will be new information that the team must cover, learn, and become familiar with. There is research involved with the early stages of this project, to learn how all of the relevant software works, and to become familiar with working in an oil and gas environment.
would have to be separated into rich and lean components for injection and lift respectively. Before running the necessary simulations to determine the compositions of each, it was first necessary to test the potential of injecting the produced gas as-is by conducting a MMP test. The MMP test in PVTSim uses two fluids and finds a minimum pressure between the two, which will allow them to be miscible. This simply means they will mix to form one homogenous solution, allowing for much easier oil production. To conduct an MMP test, the two fluids in question must be modeled in PVTSim by specifying the chemical compositions of each along with the physical properties of each chemical component. For an MMP test to work, these physical properties must be the same for both fluids. For example, the molecular weight of methane must be the same in the oil and the gas (as it is in real practice). This requirement is what raised the first issue with conduction an MMP test between the produced gas and oil. The compositions of our reservoir fluids are defined in detail up to a certain point. For carbon chains with 10 or more carbon atoms, a generalized grouping is used called the C10+ molecule. This component is used to represent all the large carbon chains by using one set of averaged physical properties (i.e. mol. weight) rather than individually listing each components properties. It is clear that this C10+ component will be different for each fluid. This poses a problem since it is required for each listed component to have the same physical properties in order to run a MMP test. Our reservoir gas and oil differ in the properties of their respective C10+ molecules. To make these two fluids compatible, a recombination test is required. This is also done using PVTSim.
gas would be more effective for injection, a separator test is required on the recombined fluid.
needed to create an accurate model. In terms of separator tests, it was necessary to conduct nine tests which used a high pressure low temperature, low pressure high temperature, mid-range pressure mid-range temperature and so on. By doing this, the group was able to determine the first and multi-contact miscibility pressures between the rich gas and reservoir fluid at the chosen temperature and pressure combinations. The separator and miscibility tests are shown on the following page in table 1. These results can show us right away that first contact miscibility is not a likely outcome using these fluids. The closest first contact miscibility pressure shown is 626.03 bar which is still more than 300 bar above our reservoir pressure of 298 bar. However, the same entry shows a value of 300 bar for multi-contact miscibility which is very close to reservoir pressure. This is a very good sign at this stage but the Design Expert analysis will show more exact miscibility values.
High/Low Test LL LH HL HH MM ML MH LM HM A B C D E F G H I Pressure (Bar) 5 5 100 100 52.5 52.5 52.5 5 100 Temperature (Deg C) 0 200 0 200 100 0 200 100 100 P sat P Crit (bar) (bar) MMP fc 539.63 539.63 539.63 539.63 539.63 539.63 539.63 539.63 539.63 577.33 727.83 631.78 670.55 634.37 590.41 674.97 699.88 642.92 626.03 762 882.71 769.96 655.9 899.2 992.4 794.73 MMP mc Drive Type
300.17 57.17% 392.14 70.50% 476.16 82.28% 396.8 71.34% 319.99 60.22% 487.36 83.26% 561.89 92.39% 413.89 74.00%
The above response surface was modeled using the minimization function in Design Expert. This function uses inputs, creates a model equation for a surface, and generates a peak where the minimum values occur on the response surface. This shows us that low pressure and low temperature are where the minimum values lie, and in this particular case, where the most important values lie the pressures which are the closest to the reservoir pressure. MUNRS then decided to refine the scope of the response modeling, by using smaller temperature and pressure ranges for the first stage of the separator. The group chose to vary the temperature from 0 to 15 degrees Celsius, and to vary the pressure from 1 to 15bar. Running separator and MMP tests again in PVTsim (with the smaller ranges of T, P) completed the refined table of values. The refined table is shown below in table 2
High/Low LL LH HL HH MM ML MH LM HM
Test A B C D E F G H I
P sat (bar) P Crit (bar) MMP fc MMP mc Drive Type 539.63 678.52 909.45 501.26 86.05% 539.63 698.27 980.35 563.06 93.86% 539.63 551.04 571.5 262.78 51.77% 539.63 575.88 622.8 297.8 56.94% 539.63 588.04 650.59 316.75 59.55% 539.63 572.1 615.15 292.8 56.08% 539.63 602.36 686.39 340.95 63.00% 539.63 689.67 948.69 534.13 90.27% 539.63 563.29 596.22 279.77 54.30%
The above values were then used as inputs into the regression modeling software, and a new model was developed for the relationship between first-stage separator temperature, pressure, and second-stage output MMP. It was found that the output MMPs were in a much more favourable range when MUNRS ran the refined model. In the earlier model, the smallest values were the only ones that were close to the target values. In this case, there were values on either side of the target reservoir pressure, so MUNRS made use of the target function in the surface-modeling component of Design Expert. By doing so, the target pressure of 298bar was selected, and the surface displaying which temperatures and pressures gave the group that target MMP is shown in figure 2.
The clear peak at several values for pressure and temperature give us a clear indication that there are possible values of temperature and pressure that correspond to useful MMPs. From the refined table, there is one value in particular that provides us with the ideal MMP: 10bar, and 15 degrees Celsius. This value gave us an MMP of 289.7bar. Since this is ideal, MUNRS will be using these separator inputs, along with the above justification of doing so (provided by Design Expert) moving forward. By selecting that temperature and pressure as the input parameters for our separator, MUNRS obtained a unique fluid composition for the rich gas exiting the second stage. That composition is shown in appendix B.
computed from this data, which relates the reservoir pressure (psig) to the oil recovery rate (MMscf/day).
In order to take this information further, an estimation of the well injection flow rate has to be completed. Using a given set of boundary conditions (reservoir pressure and well head flowing pressure) the well flow rate is determined as being the intersection between the inflow performance relation and the tubing response curves. The vertical lift performance (VLP) or lift curves can now be plotted. Using the Petroleum experts 2 VLP correlation along with our PVTsim and IPR data, a lift curve is plotted. For our purpose a 3 variable system was modeled based on Gas Oil Ratio (GOR), Top Nope Pressure and Water Cut.
The lift curves generated under these conditions gives promising results as the shape of the graph represents the desired J shape. Analyzing this VLP graph we can see that our pervious predictions based on top node pressure and GOR were correct. We noticed that as the top node or first separator pressure dropped, there was in increase in oil production. Similar to our predictions, as the oil to gas rate increased a greater injected pressure was needed to maintain the same oil recovery rate when all other variables are kept constant. It should also be noted that as the water cut was increased on the model the production decreased when all other variables are kept constant, this was also expected. From the above plot, it is determined that the use of injected gas for artificial lift does enhance oil recovery. The next step would be to determine a range of gas
injection rates and pressures based on the ECLIPSETM model to optimize the performance of the producing well.
ECLIPSETM Reservoir simulation software. The block is completely homogenous, containing uniform properties throughout. These are rock properties, like porosity and permeability. Porosity is the percentage of free space in the porous reservoir rock. Permeability is measures the ease of fluid flow through a porous media. Permeability is defined in the X, Y, and Z directions. The constant properties of the simulation are displayed in the chart below.
The fluid properties of the reservoir are taken from the donated data of the Norne field. The detailed fluid characterization contains a breakdown of the molar percentage of each molecule in the reservoir fluid. The higher carbon chains with smaller molar percentage are combined into a range because they do not represent a large percentage of the total compositions. The characterized reservoir fluid that is used in the block model is shown below.
Component N2 CO2 C1 C2 C3 iC4 nC4 iC5 nC5 C6 C7 C8 C9 C10-C12 C13-C15 C16-C18 C19-C21 C22-C24
Reservoir Fluid Molar Percentage Molecular Weight 0.135 28.014 1.116 44.01 72.088 16.043 3.892 30.07 1.685 44.097 0.318 58.124 0.593 58.124 0.262 72.151 0.277 72.151 0.473 86.178 1.259 96 1.968 107 1.419 121 3.489 146.512 2.652 189.39 2.015 235.783 1.532 275.483 1.164 317.208
The simulation is now completely defined and can soon produce results. Before initial results can be generated there is a debugging process that takes place. Based on the information being used in the simulation, the ECLIPSETM software may require addition information. A second piece of software called PVTiTM uses the fluid composition that was previously defined to identify and generate anything else ECLIPSETM may need. This completed simulation, which will now be referred to as the Basecase will serve as the reference case that all improvements will be benchmarked against. Moving forward the Basecase will evolve to incorporate some more details, like lift curves generated in Prosper; but the structural and fluid properties are now set. The Basecase simulation has been run and the results are as expected. The addition of a fully defined fluid composition increased the run time of the simulation by almost 100 times its previous duration. The completed block model, operating midway through the simulation process is shown in Figure 3. The photo shows the gas injector in the right is seen pushing oil towards to the producer in the top left.
Moving forward, the fluid compositions of the separated gas from PVTsim will be used to define a new fluid for injection and lift in the simulation. Currently the simulation uses a fluid in which no separation of light and rich components has taken place, as is industry standard today.
During the model creation process it has been observed that the addition of defined fluid composition adds a considerable amount of time to the simulation troubleshooting process. The full field Norne simulation model requires a substantial amount of work before it is ready to accept these changes. The simple block simulation will remain the top priority until usable results can be generated in all test cases, only then will focus shift to using the full field model. The omission of a full field model from this project does not reduce the validity or accuracy of the
results, but because of the time constraints it is possible that part of the scope will not be examined in this study.
necessary for our project, and that conversion was included in the original scope. It was determined that converting the code would require more time and resources that the team had at their disposal, so our ECLIPSETM modeling will be done using a block base case ECLIPSETM 300 simulation instead. This is very useful, because it can provide us with how effective our technology is in a general reservoir, which is in essence the goal of the project. Our revised Gantt chart is shown in Appendix A.
References:
"Oil and Gas Separator." Schlumberger Oilfield Glossary. N.p., 2014. Web. 05 Mar. 2014. Nangacovie, Helena L M. Application of WAG and SWAG Injection Techniques in Norne E-Segment. Thesis. Norwegian University of Science and Technology, 2012. N.p.: n.p., n.d. Print.
Appendix A
ID 1
Task Mode
Task Name
Duration
Start
Finish
Predecessors
12 Jan '14 19 Jan '14 26 Jan '14 02 Feb '14 09 Feb '14 16 Feb '14 23 Feb '14 02 Mar '14 09 Mar '14 16 Mar '14 23 Mar '14 30 Mar '14 06 Apr '14 MTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S SMTWT F S S
PVT Simulation and 41 days Gas Characterization Preliminary Research Fluid Characterization MMP Simulation 6 days
35 days
35 days
Separator Tests
35 days
Design Expert Optimization Prosper Prosper Gas Lift Simulation Eclipse Create Block Model In Eclipse 300 Reservoir Model (Near Field) Run Base Case Simulation Base Case Optimization
7 days
Mon 03/03/14
Tue 11/03/14
7 8
22 days 22 days
9 10
40 days 12 days
Mon 10/02/14 Fri 04/04/14 Mon 10/02/14 Wed 26/02/14 Wed 26/02/14 Mon 03/03/14 Wed 26/02/14 Mon 17/03/14 Tue 25/02/14 Fri 14/03/14
11
13 days
12
5 days
13
7 days
14
Run Modified 13 days Model Simulation Reservoir Model (Norne Field) Simulation Run Base Case Simulation 15 days
15
Fri 04/04/14 11
16
8 days
17
18
28 days
Fri 04/04/14
Page 1
Appendix B
Appendix C
Data used in ProsperTM : Total GOR Reservoir Temperture Reservoir Pressure Water Formation Volume Facter Oil Formation Volume Factor Oil Viscosity Gas Formation Volume Factor Oil Density Gas Density Tubing Depth Casing Depth Bubble Point Dietz Factor Drainage Area Wellbore Radius Mechanical skin Overall Heat Transfer Coefficent Cp Oil CP Gas Cp Water Top Node Pressure 225 208.94 F 4307.6 Psi 1.0328 1.32 0.318 0.0047 859.5 0.8545 2582 2692 251 31.6 340 0.354 +2 8 0.53 0.51 1 250 Rm3/Sm3 Rm3/Sm3 Cp Rm3/Sm3 kg/m3 kg/m3 m m Bara
ft BTU/h/ft2/F
psi