Hrishikesh
Lahudkar

Rotor drones can only fly for minutes inside tanks, pipelines and test chambers, and they aren't safe near people. I designed a 57 g helium blimp that moves by changing its own buoyancy instead. A micro-pump fills a 50 mL ballonet to make it heavier and pitch it nose-down; venting does the opposite, so it glides in a sawtooth path. Two nose-to-tail lines on a single servo steer it.

The envelope is a GNVR shape, 1008 mm long and 336 mm in diameter, made from eight heat-sealed gores of 15 µm Mylar. Its 59.3 L of helium lifts 61.3 g, a 4.3 g margin over the vehicle's mass.

In Abaqus, using about 32,700 shell elements, the envelope at 200 Pa averaged 1.63 MPa, within 25% of the 2.24 MPa hoop-stress hand calculation and a factor of safety of about 54. The steering-line anchor was the critical case: a 10 N pull gave a 51.86 MPa local peak and a factor of safety of only 2.3, so I specified bonded reinforcement doublers at every anchor.

A six-state longitudinal model in MATLAB, integrated with fourth-order Runge-Kutta, settled into a stable sawtooth cycle: ±6.4° of pitch at an average 35.8 cm/s, covering 204 m in ten minutes. The duty-cycled energy budget gives about 7.4 hours on a 100 mAh battery.

The prototype is built, and a first ceiling-suspended test reached the commanded ±10° of roll. Free flight is the next step.

ToolsAbaqus, MATLAB, ANSYS Fluent, CAD, heat-sealed Mylar fabrication

Rotor drones can only fly for minutes inside tanks, pipelines and test chambers, and they aren't safe near people. In nuclear facilities they add further risks: rotor downwash can stir up contaminated dust, and a crash can damage equipment in areas people can't easily enter. I designed a 57 g helium blimp that moves by changing its own buoyancy instead. A micro-pump fills a 50 mL ballonet to make it heavier and pitch it nose-down; venting does the opposite, so it glides in a sawtooth path. Two nose-to-tail lines on a single servo steer it.

The envelope is a GNVR shape, 1008 mm long and 336 mm in diameter, made from eight heat-sealed gores of 15 µm Mylar. Its 59.3 L of helium lifts 61.3 g, a 4.3 g margin over the vehicle's mass.

In Abaqus, using about 32,700 shell elements, the envelope at 200 Pa averaged 1.63 MPa, within 25% of the 2.24 MPa hoop-stress hand calculation and a factor of safety of about 54. The steering-line anchor was the critical case: a 10 N pull gave a 51.86 MPa local peak and a factor of safety of only 2.3, so I specified bonded reinforcement doublers at every anchor.

A six-state longitudinal model in MATLAB, integrated with fourth-order Runge-Kutta, settled into a stable sawtooth cycle: ±6.4° of pitch at an average 35.8 cm/s, covering 204 m in ten minutes. The duty-cycled energy budget gives about 7.4 hours on a 100 mAh battery.

The prototype is built, and a first ceiling-suspended test reached the commanded ±10° of roll. Free flight is the next step. Because it flies slowly, quietly and without propellers for hours at a time, the platform could suit indoor nuclear inspection, such as reactor halls and decommissioning sites, where it would keep people out of radiation areas.

ToolsMomentum theory, MotoCalc, power and mass budgets, bill of materials

The 2024 brief was urban air mobility: an aircraft under a 5-foot span flying timed humanitarian missions with passengers, a patient gurney and a medical cabinet, scored on speed, payload and the design report.

I led aerodynamics and stability for the 22-member team. We screened candidate aerofoils in XFLR5 at Reynolds numbers up to 430,000, shortlisted SD7062, GOE 387 and GOE 430, and chose GOE 430 for its gentle stall and high maximum lift. Static margins were set for each mission, between 16.5% and 20.5%.

Alongside the analysis I worked on the design in CATIA, SolidWorks and AutoCAD and coordinated manufacture and flight testing: nine flight attempts across two prototypes. Our design report scored 89.43 and ranked seventh in the world.

ToolsXFLR5, AVL, CATIA, SolidWorks, AutoCAD

The bracket cell makes titanium and nickel-alloy brackets for aircraft engines. I ran a DMAIC study with time studies on four operations, separating value-added work from waiting, searching and handling.

The changes were practical: 5S storage so sheets are sorted by material and thickness, standardised cooling times between welds, a foot-operated air gun for faster part changeover, and gauges kept at the point of use. Cycle time fell on every operation studied.

I also mapped the current-state value stream for a product family and proposed a future state projected to cut lead time from 42 to 28 days and lift the value-added share of work from 60% to 75%. Alongside this, I supported shop-floor non-conformance investigations to AS9100.

ToolsTime study, DMAIC, value-stream mapping, 5S, Westinghouse rating, GD&T

A vascular stent has to pass bench tests and then expand inside an artery without springing back. We modelled both in Abaqus/Standard as nonlinear static analyses with contact.

For the bench test, a half-symmetry model crushed the stent between rigid plates to 50% of its diameter, meshed with C3D10H quadratic tetrahedra. A mesh study across four element sizes converged to within 0.41% in reaction force. With an elastic-plastic material, the full stent needed about 1.13 N. A purely elastic model predicted 14.24 N, more than twelve times stiffer, with peak stresses of several gigapascals, which shows why plasticity has to be modelled.

For deployment, we placed the stent inside an idealised artery, modelled with a third-order Ogden hyperelastic law, and expanded it with a balloon of membrane elements. After inflation and deflation the stent held its shape with 1.72% recoil. Changing the stent's yield stress by ±20% moved recoil between 1.3% and 2.1%, and a second study tested how the vessel's stresses depend on its stiffness.

The 3D viewer below uses the meshes from our Abaqus input files. Its motion is a schematic based on the load steps and reported results.

ToolsAbaqus/CAE 2023, nonlinear static analysis, contact, elastic-plastic and hyperelastic materials, mesh convergence