Scribd
Upload
46 views14 pages

Winter 2016

Exam 2016

Uploaded by

Hermann Muriel
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
46 views14 pages

Winter 2016

Exam 2016

Uploaded by

Hermann Muriel
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 14
u University of Stavanger FACULTY OF SCIENCE AND TECHNOLOGY SUBJECT: PET 510 — Computational Reservoir and Well Modeling DATE: 23 February, 2016 TIME: 4 hours AID: Basic calculator is allowed THE EXAM CONSISTS OF 6 PROBLEMS ON 6 PAGES AND APPENDIX A - D REMARKS: ‘You may answer in English or Norwegian. Exercises 1 and 2 (part A) and exercises 3-6 (part B) are given equal weight. ‘COURSE RESPONSIBLE: Steinar Evje and Kjell-Kare Fjelde Problem 1. We consider a transport equation of the form (BI) uy + uy = B(x, t,u), 2 € (—00, +00) (B2) u(z,t) = uo(w) = exp(—2?) (a) Compute exact solution when b(sr t,u) - make a sketch of typical characteristics in the x — t coordinate system - verify that the computed solution satisfies (B1) and (B2) - consider your solution with the initial data given in (2) and identify the term in the solution that is responsible for the effect from the source term —u (b) We now consider the model (B1) with (2, t,u) = 0. Consider a discretization of the spatial domain [—5,5]. Assume that the domain is divided into 2M cells. It is assumed that uj = ujjj! = 0. Formulate a stable discrete scheme for cells 2,...,2M —1. |. Show that the solution u of (©) Consider (B1) again where the source term b(2,¢, u) (B1) and (B2) satisties the relation L ula, t)de [. tuo(2)dz, if we assume that u(z,t) —> 0 as 2 -+ oo. (4) Consider (B1) and (B2) with (2, t,u) = —u-+ 2. Compute exact solution and verify that it satisfies (B1) and (B2) Hint: The solution of an ODE of the form dv to [ ala)etds. fb Btn ae)—isgivenby — v(t) Problem 2. (a) Consider the pressure equation (Al) eu =tsr, 2 ER = (—00, +00) with iil data co cutyefO #<% t=o=wty= {0 250 Find an expression for the exact solution of (A1) and (A2) by using the fact that the solution with ¢ = 1 is given by (42) ul (43) Make use of an appropriate rescaling of the time variable ¢ to get £ that depends on «. Derive equation for u(z,#) where v(z,é) = u(s, t). (b) The task is the same as in (a) but make use of an appropriate rescaling of space vari- able 2 to get @ that depends on €. Derive equation for u(#,t) and make use of (A3) to get the solution of (Al) with initial data (A2). ~ Ilustrate the effect of € on the solution by making a rough sketch of the solution 1 © @ ) u(c,t) for a fixed time ¢ and a ”small” value of ¢ and a large” value of ¢. In order to derive the solution (A3) of (A1) with initial data (A2) for = 1, the idea ‘that is used is to relate the time variable t and the space variable «in a special way. Introduce a variable y that describes this relation and a variable v(y) and rewrite ‘the PDE (partial differential equation) uw = tze with initial data (A2) as and ODE, (ordinary differential equation) in terms of u(y). Find conditions for v at y = +00. Consider = tas +9(u), 2 € (0,1) u(0,t) = u(,t) =0 u(e,t = 0) = uo(2) We want to derive an upper bound for the quantity fj u®dz. - Find a condition on g(u) which will ensure that fy wde < fo ued. = In particular, give an example of a g(x) that satisfies this condition Now, consider a pressure equation (obtained by combing the mass balance equation ‘with Darcy’s law) which takes the following form k dole ep) (+) Pe= HPs, FE (0,1), P(0,t) = pli, t) =1 P(a,t = polz) where k, pi do, ¢ and ¢, are constants (permeability, viscosity, porosity, and com- pressibility constants) with initial data 1-25, o 0 in one and the same figure. - Assume that £ Make a sketch of the corresponding initial fluid velocity uo(z) and u(z,¢) and explain the motion of fluid and the corresponding change in pressure. Refer to relevant equations when you explain. - What will be the stationary solution (the solution as time goes to infinity) for pressure p and velocity u? Exam Part B - Solving Nonlinear Equations & Modelling of Well Flow There are 11 questions in total. Some formulas, equations and Matlab codes are found in Appendixes. This part constitutes 50 % of exam. Exercise 3 — Bisection method a) We are given the function f(x) = the following table. “ — 3x. Show how the bisection method works by filling out eration [xd Foe) 12) 163) 1 1 0.28172 [1,38906 | -0,01831 2 3 a b)_ Use the iterative method with a starting point xo= 0.8. Explain what happens! Exercise 4 ~ Matlab issues a) Explain what is the difference between a script file and a function file in Matlab and explain how Information is transferred between the files, b) The following is a description of the friction model we have used for two phase flow calculations: Based on this, write the matlab code for a function file that calculates the frictional gradient value based on input variables. “The friction model presented here is for a Newtonian fluids like water. It is a model that has been extended to two-phase flow by using mixture variables. The general expression for the frictional pressure loss gradient term is given by: AP ie _ 24Pr¥ nn AdS Vm) Az ou =F) (Pa/m) 17 One has to distinguish between laminar and turbulent flow and this is done by using the Reynolds number. This is defined by: Pub (0 nc )(das ~ 4) Hove Re For a Reynolds number less than 2000, the flow is considered laminar and the friction factor is given by: f= a. For Reynolds number larger than 3000, the flow is considered turbulent and the c {following expression is used: f = 0.052Re~"”. itis very important to ensure a smooth transition {for the expression of the friction factor in the transition zone (2000 < Re < 3000 ). The value for the {friction factor must be continuous as function of the Reynolds number”. Exercise 5 - Well pressures ‘We are considering a vertical well that is 5000 meter deep in total measured from rigfloor. The BOP is situated at the sea bottom with a riser connecting it to surface. The riser has an inner diameter of 119 inches. The distance from the rig floor to the BOP is 1000 meters. The airgap between the rigfloor and the sea surface is 25 meters. A9 5/8" casing with inner diameter 8.5 inches has been set to 4000 ‘mand it is hanging in the wellhead just below the BOP. We have continued drilling an 8 ” hole. The drillpipe has an outer diameter of 5.5 inches and runs from the rig down to bottom of the well. @) Ifthe flowrate is 2000 Ipm, what will be the value of the minimum liquid velocity occurring in the well? b) The mudweight measured at surface is 1.7 sg. The well is pressure dominated and we expect that this will change the effective downhole mudweight by 0.02 sg. When circulating, the friction in the annulus is 25 bars. Calculate the effective ECD (in sg) at the bottom of the well while circulating! ©) Inan emergency case, one has to disconnect the riser and move the rig away. Then we will get seawater (1.03 sg) instead of mud above the BOP (which is not closed in our calculations). What should the density of the mud remaining in the well be if it shall balance a pore pressure of 1.65 ‘sg at S000 meters? (This requirement is called to include riser margin when planning the mudweight) Dual gradient drilling is based on having two fluid densities in the annulus which can give a well pressure better fitted to the pore and fracture pressure window. Here we consider mud and seawater. We will here consider a U tube arrangement where a valve separating the drillpipe and annulus suddenly breaks and the system try to establish hydrostatic pressure equilibrium between the two connected columns. Mud will flow from drillpipe and push the seawater region further up in the annulus. Air will be sucked in on top of the drillpipe (see figure on next page). We will assume that annulus and drllpipe both have the same cross sectional area, d) Why can we put up the equation: fy, +), =h+ hy inthiscase? i> vod e) Show how we can derive the following formula PPrne = Poe) * he Pane Pree ~ Pow annulus —OrlPipe ha = e Air Exercise 6 ~ Conservation laws a) What can be said about the mass flow rate in a well in a steady state situation ? b) Describe briefly the solution technique used for solving the steady state two-phase flow conservation equations for a discretized well! Appendix A - Some Units & Formulas 1 inch =2.54 em = 0.0254 m 1 feet = 0.3048 m 1 bar = 100000 Pa 1sg=1ke/l (sg- specific gravity) M=Q-p 'M massrate (ka/s), Q Volumerate (m"/s), density (kg/m?) Q=v-4 QVolumerate (m*/s), v velocity m/s. A area m? P= p+h-0.0981 (bar), density (sa), h—vertical depth (m) Pv from Ideal gas law, NBT is in Kelvi and the relation to Celsius is K="C + 273,15 T P-V =C , Boyles law (temperature is assumed constant) Appendix B Main.m 8 Main program that calle up a routine that uses the bisection @ method to find a solution to the problem f(x) = 0. @ The search interval [a,b] is specified in the main program. @ The main program calls upon the function bisection which again calls upon 4 the function func. if error = 1, the search intervall has to be adjusted to ensure la) x £(b)<0 Specify search interval, a and b will be sent into the function bisection a= 4.0, b= 5.0; Call upon function bisection which returns the results in the variables 2 solution and error. [solution,error] = bisection(a,b)+ solution & Write to screen. error # Write to screen. Bisection.m function [solution,error] = bisection(a,b) % The numerical solver implemented here for solving the equation f(x)= 0 & 4s called Method of Halving the Interval (Bisection Method) % You will not find exact match for f(x)= 0. Maybe £(x) = 0.0001 in the end. & By using ftol we say that if abs(f(x))=0 error = 1; solution = 0; else % start iterating, we are now on the track. x3 = (x14x2)/2.0; 3 = func (x3); while (£3>ftol | £3 < -£tol) noit = noit +1; if (3¥81) <0 x2 = x37 (142) /2.05 func (x3) func (x1); end error = 0, solution = x3; noit § This statement without ; writes out the number of iterations to the screen. end funem function £ = func(x) f= x2 127 Appendix @ Program where the Larsen Cuttings Transport Model is implemented 8 First specify all input parameter: do = 8.5; % Outerdiameter (in) (1 in = 0.0254 m) di = 5; % Innerdiameter (in) rop = 33 $ Rate of Penetration - ROP ft/hr (1 ft = 0.3048m) pv = 15 $ Plastic viscosity (cP) yp = 16 % Yield point (1b£/100ft2) dcutt = 0.1 8 Cuttings diameter (in) (1 inch = 0.0254 m) mw = 10,833 @ Mudweight (ppg - pounds per gallon) 1 ppg = 119.83 kg/m3. rpm = 80 % rounds per minute cdens = 19 § cuttings density (ppg - pounds per gallon) angstart = 50 # Angle with the vertical veut - Cuttings Transport Velocity (CTF in Larsens paper) verit - Critical Transport fluid velocity (CTV) in Larsens paper. This is the minimum Eluid velocity required to maintain a continously upward movenent of the cuttings. velip - Equivalent slip velocity (ESV) defined as the velocity difference between the cuttings and the drilling fluid verit = veuttvslip All velocities are in ft/s. ua ~ apparent viscosity It should be noted that the problem is nested. Vcrit depends on vslip which again depends on an updated/correct value for verit. An iterative 4 approch on the form x(n#1) = g(x(n)) will be used. for i= 1:8 ang (i)=angstart+i*5 veut = 1/{(1-(di/do) *2)*(0.64418.16/rop) ) 7 vslipguess = 3 verit = veut + vslipguess; % Find the apparent viscosity (which depends on the "guess" for vcrit) ua = pvt (5¥yp* (do-di) } /verit % Find vslip based on the "guessed apparent viscosity". This needs to be % updated until a stable value is obtained. "Iterative approach if (ua < 53) vslip = 0.0081*ua+3. 006; else vslip = 0-02554* (ua~53)+3.287 end Now we have two estimates for vslip that can be compared and updated in a 8 while loop. The loop will end when the vslip(n#1) and vslip (n) do not @ change much anymore. I.e the iterative solution is found. while (abs (vslip-vslipguess) }>0.01 vslipguess = vslip; verit = veut + vslipguess; @ Find the apparent viscosity (which depends on the "guess" for vcrit) ua = py+ (5*yp* (do-di)) /verity % Find vslip based on the "guessed apparent viscosity". This needs to be ® updated until a stable value is obtained. "Iterative approach". if (ua < 53) velip = 0.0051*ua+3.006; else vslip = 0.02554* (ua-53)+3.28; end nentl; vslip % Take away ; and you will se how vslip converges to a solution end § End while loop 8 % Cuttings size correction factor: CZ = ~1.05¥dcutt+1.286 % Mud Weight Correction factor (Buoancy effect) Af (mw>8.7) CMW = 1-0,0333* (mw-8.7) else cum = 1.0 end -05D50cut+1.286 § Angle correction factor ANG = 0,0342*ang (4) -0.000233*ang(i)*2-0.213 vslip = vslip*CZ¥CMi*CANG; & Inelude correction factors. 8 Find final minimum velocity required for cuttings transport (ft/s) vorit = veut + velip veritms = verit*0.3048 % Velocity in m/s Q = 3.14/4*( (8.5*0.0254) *2~(5*0.0254) *2) *vorLtms & (m3/s) Q = 9*60*1000 & (1pm) yrate(i)=9 end plot (ang, yrate) Appendix D - Steady State Model for Two Phase Flow Conservation of liquid mass a 2 (pay,)=0 a Pav) Conservation of gas mass Zedpya,y,)=0 ez Conservation of momentum. a ap. Be P= Pe # Gas slippage model (simple): y, Kg, +S (Ke1.2) Liquid density model (simple) pil) = 9. + 2=P2), assume water: a 1, =1000kg/m?, p, =100000Pa, a, = 1500 m/s Gas density model (simple) Pp, () 4 ideal gas: a, =316 m/s. Friction model ‘The friction model presented here is for a Newtonian fluids like water. The general expression for the frictional pressure loss gradient term is given by: AP _ 24D insta) ae (ou — 4a) (Palm) A-(o*) p,~ phase densities (kg/m*), liquid —> isl, gas ->i = vy, - phase velocities (m/s) 14, ~ Phase viscosity (Pas) p- pressure (Pa) 8 —gravity constant 9.81 m/s? @, - phase volume fractions taking values between 0 and 1. a, +a, Pu = % P; +, P,~ mixture density /¥) +a,V, ~ mixture velocity Hae = fy + 01, H.- mixture viscosity

You might also like