01.06.2017 Views

UWE Bristol Engineering showcase 2015

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Heather Arnall<br />

MEng Mechanical <strong>Engineering</strong><br />

Project Supervisor<br />

Ramin Amali<br />

Repair of Filament Wound Composite Pipes<br />

Modelling the pipe repair in ABAQUS<br />

An internal pressure of 6MPa were applied to the<br />

pipes, which had one end of the pipe closed to<br />

consider both the longitudinal and hoop stress.<br />

Partitions were used to model the removed<br />

damage and wrap. The wrap has a fibre<br />

orientation of [0 90 0 90]s, whilst the pipe is [54.75<br />

-54.75]4 and glass fibre properties were applied to<br />

both. An intensive mesh study found that a global<br />

mesh of 7mm was required to be applied to satisfy<br />

nominal stress. The region surrounding the hole<br />

was refined with 15 elements per line and four<br />

additional lines were applied to structure the<br />

mesh, providing a mesh of 8,860 elements.<br />

Pipe parameter investigations<br />

Three different pipe diameters (0.1m, 0.3m and<br />

0.5m) were considered to determine the FOS of<br />

the pipes when undamaged, damaged and<br />

repaired. The FOS damage reduction factor, FOS<br />

wrap gain factor and FOS overall reduction factor<br />

were then obtained.<br />

FFFFFF rrrrrrrrrrrrrrrrrr oooo gggggggg ffffffffffff ff = FFFFFF pppppppppp<br />

FFFFFF aaaaaaaaaa<br />

Formulas were created through plotting each of<br />

these factors against pipe diameter. Two additional<br />

diameters (0.08m and 0.2m) were analysed to<br />

determine the formulas have an accuracy of 95%.<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr =<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

Glass fibre Air Units<br />

Poisson’s Ratio 0.3 0.03 -<br />

Young’s Modulus - 0.01 MPa<br />

long. modulus E1 120 - GPa<br />

trans. modulus E2 8 - GPa<br />

shear modulus 6 - GPa<br />

long. tension Xt 1800 - MPa<br />

long. comp. Xc -1200 - MPa<br />

trans.tension Yt 80 - MPa<br />

trans. comp. Yc -200 - MPa<br />

shear S 150 - MPa<br />

FFFFFF pppppppppp dddddddddddd<br />

−11.69DD 2 + 9.016DD + 1.1013<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr<br />

3.1873DD 2 − 2.1499DD + 0.708<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

FFFFFF pppppppppp dddddddddddd<br />

2.7236DD 2 − 1.2208DD + 1.0848<br />

Formulas for pipe thickness for thin walled<br />

pipes were obtained through the same process for<br />

three pipe thicknesses: 3mm, 4mm and 5mm. Pipe<br />

thicknesses of 2mm and 3.5mm were modeled to<br />

determine formula accuracy of 99% for following<br />

repair, whilst 66% for following removal. The<br />

overall FOS reduction factor was constant for<br />

varying thicknesses, this was also confirmed<br />

through modelling a 0.3m diameter pipe with<br />

thicknesses of 3mm and 15mm.<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr =<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

FFFFFF PPPPPPPPPP dddddddddddd<br />

0.0174TT 2 − 0.178TT + 2.1234<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr<br />

−0.0066TT 2 + 0.0669TT + 0.449<br />

FFFFFF pppppppppp dddddddddddd<br />

−0.0012TT 2 + 0.01TT + 1.01TT + 1.0109<br />

Investigation into damage removal<br />

Damage should be removed with a circular cut<br />

out, but this is not always possible. Three cut out<br />

areas were considered, finding that using a square<br />

cut out at 45 degrees reduces the FOS following<br />

damage removal by 40%, but does not significantly<br />

affect the FOS following repair compared to a<br />

circular cut out. The three circular cut outs<br />

(diameters: 30mm, 50mm and 70mm) were used<br />

to create the formulas shown. By analysing a 0.4m<br />

damage diameter it was found that the accuracy<br />

for the FOS following damage removal was 72%,<br />

whilst 99.98% for following repair.<br />

FFFFFF pppppppppp dddddddddddd<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr =<br />

−222.32dd 2 + 27.23dd + 0.839<br />

FFFFFF aaaaaaaaaa dddddddddddd rrrrrrrrrrrrrr<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

117.05dd 2 − 14.232dd + 1.1695<br />

FFFFFF pppppppppp dddddddddddd<br />

FFFFFF aaaaaaaaaa rrrrrrrrrrrr =<br />

2.6216dd 2 − 0.2097dd + 1.2382<br />

Wrap Width W (mm)<br />

Investigation into wrap dimensions<br />

Nine different wrap thicknesses were considered<br />

between 2.4mm and 28mm and as expected as<br />

the wrap thickness increased, so did the FOS<br />

following repair. Three different wrap thicknesses<br />

(4, 8 and 13mm) and three different damage sizes<br />

(0.3m, 0.5m and 0.7m) were then used for an<br />

extensive trial and error method to determine the<br />

required wrap width for each scenario to obtain<br />

the following graph.<br />

Wrap width for damage diameters less than 0.07m<br />

400<br />

350<br />

300<br />

and wrap thicknesses less than 13mm<br />

W = 100000d 2 - 6500d + 305<br />

250<br />

200<br />

150<br />

W = 112500d 2 - 8500d + 263.75<br />

100<br />

50<br />

0<br />

W = 75000d 2 - 5500d + 187.5<br />

0.03 0.04 0.05 0.06 0.07<br />

Damage diameter d (m)<br />

13mm Wrap thickness 8mm Wrap thickness 4mm Wrap thickness<br />

Investigation into adhesives<br />

Epoxy and Polyester test pieces were produced<br />

and tensile tested to find the failure of the<br />

polyester pieces were less predictable, at a lower<br />

stress and more catastrophic. From these<br />

experiments the young’s moduli were obtained<br />

and used to FEA model the adhesive used when<br />

repairing composite pipes (polyurethane was also<br />

considered). It was found that the adhesive<br />

material had no effect on the FOS following repair,<br />

but there was S 11 . Varying thicknesses were<br />

considered to see that when plotted against S 11<br />

there was a quadratic relationship.<br />

Project summary<br />

Composite pipes are used in a variety of industries<br />

and put under a variety of loads. If loaded too much<br />

then damage or failure could occur. Composite pipes<br />

can be repaired through the use of a composite wrap,<br />

however it is important to ensure the procedure is<br />

done correctly to ensure a good quality repair.<br />

Project Objectives<br />

The main aim of this investigation is to use FEA<br />

software to determine the effect of the parameters<br />

involved in filament wound pipe repair during both<br />

the removal of damage and the application of the<br />

wrap. Investigations will consider pipe diameter and<br />

thickness, damage size and shape, wrap thickness<br />

and optimum width and adhesive material and<br />

thickness.<br />

Project Conclusion<br />

In this project over 150 FEA models were used to<br />

investigate the effect of the repair parameters.<br />

Formulas, which will save time and ensure correct<br />

repair methods, have been created to predict the FOS<br />

following damage removal and following repair for<br />

varying pipe diameters, circular cut out sizes and<br />

optimum wrap width. It was found that if the pipe is<br />

thin walled then the pipe thickness has no effect on<br />

the FOS following repair. It was also found that<br />

removing damage with a square cut out reduces the<br />

FOS following damage removal by 40% compared to a<br />

circular cut out, but there was no significant<br />

difference in FOS once the pipes had been repaired.<br />

As expected as the wrap thickness increased the FOS<br />

following repair increased. From the adhesive study it<br />

is suggested that an Epoxy adhesive should be used in<br />

the repair of composite pipes due to a higher<br />

ultimate strength and a less catastrophic failure than<br />

polyester. FEA confirmed this by finding the least<br />

stress was caused through Epoxy, compared to<br />

Polyester and Polyurethane.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!