Actual source code: ex5.c
1: static char help[] = "Basic equation for an induction generator driven by a wind turbine.\n";
3: /*F
4: \begin{eqnarray}
5: T_w\frac{dv_w}{dt} & = & v_w - v_we \\
6: 2(H_t+H_m)\frac{ds}{dt} & = & P_w - P_e
7: \end{eqnarray}
8: F*/
9: /*
10: - Pw is the power extracted from the wind turbine given by
11: Pw = 0.5*\rho*cp*Ar*vw^3
13: - The wind speed time series is modeled using a Weibull distribution and then
14: passed through a low pass filter (with time constant T_w).
15: - v_we is the wind speed data calculated using Weibull distribution while v_w is
16: the output of the filter.
17: - P_e is assumed as constant electrical torque
19: - This example does not work with adaptive time stepping!
21: Reference:
22: Power System Modeling and Scripting - F. Milano
23: */
25: #include <petscts.h>
27: #define freq 50
28: #define ws (2 * PETSC_PI * freq)
29: #define MVAbase 100
31: typedef struct {
32: /* Parameters for wind speed model */
33: PetscInt nsamples; /* Number of wind samples */
34: PetscReal cw; /* Scale factor for Weibull distribution */
35: PetscReal kw; /* Shape factor for Weibull distribution */
36: Vec wind_data; /* Vector to hold wind speeds */
37: Vec t_wind; /* Vector to hold wind speed times */
38: PetscReal Tw; /* Filter time constant */
40: /* Wind turbine parameters */
41: PetscScalar Rt; /* Rotor radius */
42: PetscScalar Ar; /* Area swept by rotor (pi*R*R) */
43: PetscReal nGB; /* Gear box ratio */
44: PetscReal Ht; /* Turbine inertia constant */
45: PetscReal rho; /* Atmospheric pressure */
47: /* Induction generator parameters */
48: PetscInt np; /* Number of poles */
49: PetscReal Xm; /* Magnetizing reactance */
50: PetscReal Xs; /* Stator Reactance */
51: PetscReal Xr; /* Rotor reactance */
52: PetscReal Rs; /* Stator resistance */
53: PetscReal Rr; /* Rotor resistance */
54: PetscReal Hm; /* Motor inertia constant */
55: PetscReal Xp; /* Xs + Xm*Xr/(Xm + Xr) */
56: PetscScalar Te; /* Electrical Torque */
58: Mat Sol; /* Solution matrix */
59: PetscInt stepnum; /* Column number of solution matrix */
60: } AppCtx;
62: /* Initial values computed by Power flow and initialization */
63: PetscScalar s = -0.00011577790353;
64: /*Pw = 0.011064344110238; %Te*wm */
65: PetscScalar vwa = 22.317142184449754;
66: PetscReal tmax = 20.0;
68: /* Saves the solution at each time to a matrix */
69: PetscErrorCode SaveSolution(TS ts)
70: {
71: AppCtx *user;
72: Vec X;
73: PetscScalar *mat;
74: const PetscScalar *x;
75: PetscInt idx;
76: PetscReal t;
78: PetscFunctionBegin;
79: PetscCall(TSGetApplicationContext(ts, &user));
80: PetscCall(TSGetTime(ts, &t));
81: PetscCall(TSGetSolution(ts, &X));
82: idx = 3 * user->stepnum;
83: PetscCall(MatDenseGetArray(user->Sol, &mat));
84: PetscCall(VecGetArrayRead(X, &x));
85: mat[idx] = t;
86: PetscCall(PetscArraycpy(mat + idx + 1, x, 2));
87: PetscCall(MatDenseRestoreArray(user->Sol, &mat));
88: PetscCall(VecRestoreArrayRead(X, &x));
89: user->stepnum++;
90: PetscFunctionReturn(PETSC_SUCCESS);
91: }
93: /* Computes the wind speed using Weibull distribution */
94: PetscErrorCode WindSpeeds(AppCtx *user)
95: {
96: PetscScalar *x, *t, avg_dev, sum;
97: PetscInt i;
99: PetscFunctionBegin;
100: user->cw = 5;
101: user->kw = 2; /* Rayleigh distribution */
102: user->nsamples = 2000;
103: user->Tw = 0.2;
104: PetscOptionsBegin(PETSC_COMM_WORLD, NULL, "Wind Speed Options", "");
105: {
106: PetscCall(PetscOptionsReal("-cw", "", "", user->cw, &user->cw, NULL));
107: PetscCall(PetscOptionsReal("-kw", "", "", user->kw, &user->kw, NULL));
108: PetscCall(PetscOptionsInt("-nsamples", "", "", user->nsamples, &user->nsamples, NULL));
109: PetscCall(PetscOptionsReal("-Tw", "", "", user->Tw, &user->Tw, NULL));
110: }
111: PetscOptionsEnd();
112: PetscCall(VecCreate(PETSC_COMM_WORLD, &user->wind_data));
113: PetscCall(VecSetSizes(user->wind_data, PETSC_DECIDE, user->nsamples));
114: PetscCall(VecSetFromOptions(user->wind_data));
115: PetscCall(VecDuplicate(user->wind_data, &user->t_wind));
117: PetscCall(VecGetArray(user->t_wind, &t));
118: for (i = 0; i < user->nsamples; i++) t[i] = (i + 1) * tmax / user->nsamples;
119: PetscCall(VecRestoreArray(user->t_wind, &t));
121: /* Wind speed deviation = (-log(rand)/cw)^(1/kw) */
122: PetscCall(VecSetRandom(user->wind_data, NULL));
123: PetscCall(VecLog(user->wind_data));
124: PetscCall(VecScale(user->wind_data, -1 / user->cw));
125: PetscCall(VecGetArray(user->wind_data, &x));
126: for (i = 0; i < user->nsamples; i++) x[i] = PetscPowScalar(x[i], (1 / user->kw));
127: PetscCall(VecRestoreArray(user->wind_data, &x));
128: PetscCall(VecSum(user->wind_data, &sum));
129: avg_dev = sum / user->nsamples;
130: /* Wind speed (t) = (1 + wind speed deviation(t) - avg_dev)*average wind speed */
131: PetscCall(VecShift(user->wind_data, (1 - avg_dev)));
132: PetscCall(VecScale(user->wind_data, vwa));
133: PetscFunctionReturn(PETSC_SUCCESS);
134: }
136: /* Sets the parameters for wind turbine */
137: PetscErrorCode SetWindTurbineParams(AppCtx *user)
138: {
139: PetscFunctionBegin;
140: user->Rt = 35;
141: user->Ar = PETSC_PI * user->Rt * user->Rt;
142: user->nGB = 1.0 / 89.0;
143: user->rho = 1.225;
144: user->Ht = 1.5;
145: PetscFunctionReturn(PETSC_SUCCESS);
146: }
148: /* Sets the parameters for induction generator */
149: PetscErrorCode SetInductionGeneratorParams(AppCtx *user)
150: {
151: PetscFunctionBegin;
152: user->np = 4;
153: user->Xm = 3.0;
154: user->Xs = 0.1;
155: user->Xr = 0.08;
156: user->Rs = 0.01;
157: user->Rr = 0.01;
158: user->Xp = user->Xs + user->Xm * user->Xr / (user->Xm + user->Xr);
159: user->Hm = 1.0;
160: user->Te = 0.011063063063251968;
161: PetscFunctionReturn(PETSC_SUCCESS);
162: }
164: /* Computes the power extracted from wind */
165: PetscErrorCode GetWindPower(PetscScalar wm, PetscScalar vw, PetscScalar *Pw, AppCtx *user)
166: {
167: PetscScalar temp, lambda, lambda_i, cp;
169: PetscFunctionBegin;
170: temp = user->nGB * 2 * user->Rt * ws / user->np;
171: lambda = temp * wm / vw;
172: lambda_i = 1 / (1 / lambda + 0.002);
173: cp = 0.44 * (125 / lambda_i - 6.94) * PetscExpScalar(-16.5 / lambda_i);
174: *Pw = 0.5 * user->rho * cp * user->Ar * vw * vw * vw / (MVAbase * 1e6);
175: PetscFunctionReturn(PETSC_SUCCESS);
176: }
178: /*
179: Defines the ODE passed to the ODE solver
180: */
181: static PetscErrorCode IFunction(TS ts, PetscReal t, Vec U, Vec Udot, Vec F, AppCtx *user)
182: {
183: PetscScalar *f, wm, Pw, *wd;
184: const PetscScalar *u, *udot;
185: PetscInt stepnum;
187: PetscFunctionBegin;
188: PetscCall(TSGetStepNumber(ts, &stepnum));
189: /* The next three lines allow us to access the entries of the vectors directly */
190: PetscCall(VecGetArrayRead(U, &u));
191: PetscCall(VecGetArrayRead(Udot, &udot));
192: PetscCall(VecGetArray(F, &f));
193: PetscCall(VecGetArray(user->wind_data, &wd));
195: f[0] = user->Tw * udot[0] - wd[stepnum] + u[0];
196: wm = 1 - u[1];
197: PetscCall(GetWindPower(wm, u[0], &Pw, user));
198: f[1] = 2.0 * (user->Ht + user->Hm) * udot[1] - Pw / wm + user->Te;
200: PetscCall(VecRestoreArray(user->wind_data, &wd));
201: PetscCall(VecRestoreArrayRead(U, &u));
202: PetscCall(VecRestoreArrayRead(Udot, &udot));
203: PetscCall(VecRestoreArray(F, &f));
204: PetscFunctionReturn(PETSC_SUCCESS);
205: }
207: int main(int argc, char **argv)
208: {
209: TS ts; /* ODE integrator */
210: Vec U; /* solution will be stored here */
211: Mat A; /* Jacobian matrix */
212: PetscMPIInt size;
213: PetscInt n = 2, idx;
214: AppCtx user;
215: PetscScalar *u;
216: SNES snes;
217: PetscScalar *mat;
218: const PetscScalar *x, *rmat;
219: Mat B;
220: PetscScalar *amat;
221: PetscViewer viewer;
223: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
224: Initialize program
225: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
226: PetscFunctionBeginUser;
227: PetscCall(PetscInitialize(&argc, &argv, (char *)0, help));
228: PetscCallMPI(MPI_Comm_size(PETSC_COMM_WORLD, &size));
229: PetscCheck(size == 1, PETSC_COMM_WORLD, PETSC_ERR_WRONG_MPI_SIZE, "Only for sequential runs");
231: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
232: Create necessary matrix and vectors
233: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
234: PetscCall(MatCreate(PETSC_COMM_WORLD, &A));
235: PetscCall(MatSetSizes(A, n, n, PETSC_DETERMINE, PETSC_DETERMINE));
236: PetscCall(MatSetFromOptions(A));
237: PetscCall(MatSetUp(A));
239: PetscCall(MatCreateVecs(A, &U, NULL));
241: /* Create wind speed data using Weibull distribution */
242: PetscCall(WindSpeeds(&user));
243: /* Set parameters for wind turbine and induction generator */
244: PetscCall(SetWindTurbineParams(&user));
245: PetscCall(SetInductionGeneratorParams(&user));
247: PetscCall(VecGetArray(U, &u));
248: u[0] = vwa;
249: u[1] = s;
250: PetscCall(VecRestoreArray(U, &u));
252: /* Create matrix to save solutions at each time step */
253: user.stepnum = 0;
255: PetscCall(MatCreateSeqDense(PETSC_COMM_SELF, 3, 2010, NULL, &user.Sol));
257: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
258: Create timestepping solver context
259: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
260: PetscCall(TSCreate(PETSC_COMM_WORLD, &ts));
261: PetscCall(TSSetProblemType(ts, TS_NONLINEAR));
262: PetscCall(TSSetType(ts, TSBEULER));
263: PetscCall(TSSetIFunction(ts, NULL, (TSIFunction)IFunction, &user));
265: PetscCall(TSGetSNES(ts, &snes));
266: PetscCall(SNESSetJacobian(snes, A, A, SNESComputeJacobianDefault, NULL));
267: /* PetscCall(TSSetIJacobian(ts,A,A,(TSIJacobian)IJacobian,&user)); */
268: PetscCall(TSSetApplicationContext(ts, &user));
270: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
271: Set initial conditions
272: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
273: PetscCall(TSSetSolution(ts, U));
275: /* Save initial solution */
276: idx = 3 * user.stepnum;
278: PetscCall(MatDenseGetArray(user.Sol, &mat));
279: PetscCall(VecGetArrayRead(U, &x));
281: mat[idx] = 0.0;
283: PetscCall(PetscArraycpy(mat + idx + 1, x, 2));
284: PetscCall(MatDenseRestoreArray(user.Sol, &mat));
285: PetscCall(VecRestoreArrayRead(U, &x));
286: user.stepnum++;
288: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
289: Set solver options
290: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
291: PetscCall(TSSetMaxTime(ts, 20.0));
292: PetscCall(TSSetExactFinalTime(ts, TS_EXACTFINALTIME_MATCHSTEP));
293: PetscCall(TSSetTimeStep(ts, .01));
294: PetscCall(TSSetFromOptions(ts));
295: PetscCall(TSSetPostStep(ts, SaveSolution));
296: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
297: Solve nonlinear system
298: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
299: PetscCall(TSSolve(ts, U));
301: PetscCall(MatCreateSeqDense(PETSC_COMM_SELF, 3, user.stepnum, NULL, &B));
302: PetscCall(MatDenseGetArrayRead(user.Sol, &rmat));
303: PetscCall(MatDenseGetArray(B, &amat));
304: PetscCall(PetscArraycpy(amat, rmat, user.stepnum * 3));
305: PetscCall(MatDenseRestoreArray(B, &amat));
306: PetscCall(MatDenseRestoreArrayRead(user.Sol, &rmat));
308: PetscCall(PetscViewerBinaryOpen(PETSC_COMM_SELF, "out.bin", FILE_MODE_WRITE, &viewer));
309: PetscCall(MatView(B, viewer));
310: PetscCall(PetscViewerDestroy(&viewer));
311: PetscCall(MatDestroy(&user.Sol));
312: PetscCall(MatDestroy(&B));
313: /* - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
314: Free work space. All PETSc objects should be destroyed when they are no longer needed.
315: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - */
316: PetscCall(VecDestroy(&user.wind_data));
317: PetscCall(VecDestroy(&user.t_wind));
318: PetscCall(MatDestroy(&A));
319: PetscCall(VecDestroy(&U));
320: PetscCall(TSDestroy(&ts));
322: PetscCall(PetscFinalize());
323: return 0;
324: }
326: /*TEST
328: build:
329: requires: !complex
331: test:
333: TEST*/