Duct Pulsation Problem Captured and Solved Using STAR-CCM+ Eric Duplain, Eng., M.Eng. (BMA) François McKenty, Eng., Ph.D. (BMA) Normand Brais, Eng., Ph.D. (BMA) John Viskup, President (Victory Energy)
Context Stack FGR FD FAN Fresh Air Inlet Steam out (heating) Fresh Air + FGR Duct to Burner STRONG VIBRATION Boiler MAX REACHABLE LOAD 50 % Burner 2
Context Less than two months left before heating season starts MUST WORK! Fan to burner ductwork evaluated by burner manufacturer as acceptable Turnaround time: two weeks! 3
Context Fan to burner ductwork subject From FD FAN Upstairs to severe space limitation Air flow probably not optimal Although evaluated as acceptable, ductwork must be investigated To Burner & Boiler 4
Context "Gregg, I took this video with my phone. The floor is shaking and it's hard to hold still. Near the end I moved the phone over to a floor mounted brace to steady the phone. You can still see the housing moving when comparing to the concrete motor base. End-Customer 5
Context 6
Problem Analysis 8 sec 9 sec 10 sec 11 sec 12 sec 23 pulsations in 4 seconds = 5.75 Hz 7
Problem Analysis Possible causes: Problematic fan Unstable flame in boiler Tube bank resonant frequency > 1500 rpm Vibration present EVEN without combustion Aerodynamics in ducts 8
Problem Analysis FD Fan 4½ x 3½ ft 4½ x 6 ft Burner/ Boiler @ 50 % LOAD : 37 ft/s 5.75 Hz = 6.4 ft 22 ft/s 5.75 Hz = 3.8 ft 9
Problem Analysis Suspect: Eddies detaching 5.75 times per second 10
CFD Analysis - GEOMETRY Velocity Inlet : Uniform profile Expansion joint Air/FGR duct Sharp elbow Windbox Burner Flow split outlet after burner 11
CFD Analysis - MESH Polyhedral mesh Base size 5 cm ( 2 in) 3 prism layer, thickness 6 mm ( ¼ in) 550k cells 12
CFD Analysis PHYSICS STEADY STATE Code: STAR-CCM+ 7.04.006 Fluid flow: steady state Turbulence: k- model (Two-layer all y+ wall treatment) Species: air with corrected density for FGR & temperature Operating condition: 50 % LOAD Porous media to simulate burner effect on flow 13
CFD Analysis RESULTS BASE CASE STEADY STATE 50% LOAD : 14
CFD Analysis RESULTS BASE CASE STEADY STATE 50% LOAD : 15
CFD Analysis RESULTS BASE CASE STEADY STATE 50% LOAD : 16
CFD Analysis RESULTS BASE CASE STEADY STATE 50% LOAD : 17
CFD Analysis RESULTS STEADY STATE PRELIMINARY CONCLUSIONS: Large recirculation after sharp turn Will surely induce flow instability (turbulence) High residuals indicate transient phenomenon may be occuring Must complete analysis with transient CFD run 18
CFD Analysis PHYSICS TRANSIENT Code: STAR-CCM+ 7.04.006 Fluid flow: transient Time step : 1/10 th of one expected cycle = 1 5.75Hz 10 = 17 ms 5 ms chosen for safety (given time constraint) Total time: up to 10 sec. Turbulence: k- model (Two-layer all y+ wall treatment) Species: Air with corrected density for FGR & temperature Operating conditions: 50 %, 100 % LOADS Porous media to simulate burner effect on flow 19
Static Pressure [in H2O] CFD Analysis RESULTS TRANSIENT Base Case 50 % LOAD 10.5 10.3 10.1 9.9 9.7 9.5 0 1 2 3 4 5 6 7 8 9 10 Time [sec.] 20
Static Pressure [in H2O] CFD Analysis RESULTS TRANSIENT Base Case 100 % LOAD 35 34.5 34 33.5 33 32.5 32 0 1 2 3 4 5 6 7 8 9 10 Time [sec.] 21
CFD Analysis RESULTS TRANSIENT Base Case 100 % LOAD Amplitude: ± 1 in w.c. Pulsation: 4-5 Hz 50 % LOAD Amplitude: ± 0.25 in w.c. Pulsation: 2-3 Hz 22
CFD Analysis RESULTS TRANSIENT Base Case Duct side wall 5 ft x 10 ft 100 % LOAD Amplitude: ± 1 in w.c. Pulsation: 4-5 Hz 520 lbs Force Enough to make duct side wall move! 50 % LOAD Amplitude: ± 0.25 in w.c. Pulsation: 2-3 Hz 130 lbs Force 23
CFD Analysis SOLUTION GOAL: Stabilize the flow Must break the detaching eddies OPTIONS: Re-design ducting Install turning vanes Expensive and not possible given constrained schedule Solution tried: 1 Turning vane 2 Turning vanes 3 Turning vanes 3 Turning vanes + perforated plate at windbox inlet 24
CFD Analysis SOLUTION 3 turning vanes + perforated plate at burner inlet Perforated plate 1 in, 1.25 in C-C, staggered 2 in w.c. P @ FULL LOAD 3 turning vanes 25
CFD Analysis MESH SOLUTION Polyhedral mesh Base size 4.5 cm (1¾ in) Turning vanes 2.25 cm (0.9 in), Perforated plate 6 mm (¼ in) 5.5M cells (10X base case) 26
CFD Analysis RESULTS SOLUTION STEADY STATE 50 % LOAD : 27
CFD Analysis RESULTS SOLUTION STEADY STATE 50% LOAD : 28
CFD Analysis RESULTS SOLUTION STEADY STATE 50 % LOAD : 29
CFD Analysis RESULTS SOLUTION STEADY STATE 50 % LOAD : 30
Static Pressure [in H2O] CFD Analysis RESULTS SOLUTION TRANSIENT, 50 % LOAD: 10.5 10.3 10.1 9.9 9.7 9.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time [sec.] 31
Static Pressure [in H2O] CFD Analysis RESULTS SOLUTION TRANSIENT, 100 % LOAD: 35 34 33 32 31 30 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time [sec.] 32
CFD Analysis RESULTS SOLUTION TRANSIENT 100 % LOAD Amplitude: ± 0.05 in w.c. (20X smaller) Pulsation: NONE Base Case Solution 50 % LOAD Amplitude: ± 0.025 in w.c. (10X smaller) Pulsation: NONE 33
CFD Analysis RESULTS TRANSIENT Two weeks after initial email: satisfactory solution! Drawings of perforated plate and turning vanes prepared and sent Installation of parts started two days later Boiler started: 100% LOAD reached without pulsation! Problem Solved End Customer Happy 34
Conclusions Although ductwork was supplied by a third party, Victory Energy hired BMA to conduct CFD analysis Air flow instability indicated by steady state CFD runs Pulsation phenomenon captured using transient CFD runs 2-5 Hz simulated vs. 5.75 Hz measured, close enough given that: Incomplete geometry provided Exact load around 50 % but unknown Proposed solution: turning vanes and perforated plate Pulsation phenomenon eliminated for all simulated loads 35
Conclusions Total Turnaround time between 1 st phone call, CFD study and recommendations: 2 weeks 36
Hardware Machines Operating System Linux 3.0.26-0.7-default (SUSE Enterprise Server 11) CPU Type CPU Addressability CPU Count CPU Cache Physical Memory Intel(R) Xeon(R) CPU E5-1620 0 @ 3.60GHz (x86_64) 64 bit 4 (4 cores/socket, Hyper-threading) 10 240 KB (L2) 15 926 MB 10 machines x 4 cores/machine 40 cores total Network 1 Gb Ethernet Computing Time Description Type # poly cells Iter CPU time Elapsed time Base Case Steady State 553 819 1 000 5 h 54 min 9 min Solution Steady State 5 553 494 1 000 38 h 50 min 58 min Base Case Solution Transient, 10 sec., 0.005 sec. time step, 20 iter/step Transient, 10 sec., 0.005 sec. time step, 20 iter/step 553 819 40 000 236 h 6 h 9 min 5 553 494 40 000 1 553 h 38 h 55 min 37