Concept-to-drawing-to-prototype development across mechanical, electromechanical, and process systems — built to perform under real operating conditions.
By the time a drawing is released, most of the product's lifetime cost is already committed — material choice, part count, joining method, tolerance scheme, the decision to use a casting rather than a fabrication. Manufacturing and sourcing can then optimise within a few per cent of what design has fixed. This is not a controversial claim, and yet detailed engineering design is routinely treated as a downstream drafting activity rather than the point where the economics of the product are actually set.
The second recurring failure is validation designed to pass. A test protocol written after the design is complete, by the team that produced it, tends to exercise the conditions the design handles well. Field failures then arrive from the load case nobody characterised — a thermal cycling profile, a resonance at an excitation frequency the mounting scheme never anticipated, an interface pressure that relaxed over five years.
The practice provides detailed engineering design as a consultancy from Vadodara, working with manufacturers in high-voltage equipment, cable and cable accessories, railways and rolling stock, battery energy storage, process plant, oil and gas, and metals and mining, across India and with European clients.
Electromechanical design is where discipline boundaries do the most damage. The mechanical engineer sizes for structural load; the electrical engineer sizes for ampacity and dielectric margin; nobody owns the coupled problem, which is usually thermal.
The practice designs across that coupling. On a current-carrying assembly, conductor cross-section is not set by ampacity alone but by temperature-rise limits at the contact interface, by the short-circuit electromagnetic force the supports must react — which scales with the square of peak current and inversely with conductor spacing — and by the thermal expansion the joint design must accommodate without losing contact pressure. On insulation systems, creepage and clearance follow from pollution degree and comparative tracking index of the material actually specified, and field grading is checked against the geometry as manufactured, not as idealised, because the triple point at an interface is where partial discharge starts.
Simulation is used where it changes a decision. Structural and thermal FEA, coupled electromagnetic-thermal analysis, CFD for flow and thermal management, and multi-body dynamics where kinematics dominate. Model credibility is treated as part of the deliverable: mesh convergence demonstrated rather than asserted, contact and boundary conditions justified, material data sourced and stated, and results reconciled against a hand calculation or a measured datum wherever one exists. A model that has not been correlated to anything is an opinion with contour plots.
Process equipment design carries code obligations that constrain the design space before optimisation begins. Pressure equipment is designed and verified to the applicable code — ASME Section VIII Division 1 or Division 2, or EN 13445 where European conformity is required — with nozzle reinforcement, local load and fatigue assessment treated as design activities rather than as certification paperwork. Storage tanks, structural steelwork, piping and support systems follow the same logic: the code sets the floor, the duty cycle sets the real requirement.
Fatigue is the most commonly under-treated of these. Equipment specified for a static design pressure often sees a cyclic duty — batch operation, pressure swing, start-stop thermal transients — and the weld detail that satisfies a static check may be a class that will not survive the cycle count. Establishing the actual load spectrum before selecting weld details and inspection categories is usually cheaper than the alternative.
DFM is only useful when it is specific to the plant that will build the part. Generic guidelines about draft angles and uniform wall thickness are true and mostly already known. The value comes from designing to the tolerance capability of the machines on that shop floor, the fixturing available, the operator skill distribution, and the inspection method that will be used in production rather than in the metrology room.
Assembly work follows Boothroyd-Dewhurst logic: reduce part count by asking, for each part, whether it must move relative to its neighbour, must be a different material, or must be separable for assembly to be possible. Parts that fail all three are candidates for consolidation. Alongside that, remove the conditions that require judgement — asymmetry that is nearly symmetric, fasteners that can be started cross-threaded, orientations that can be got wrong and will not be detected until test.
Tolerance work is done quantitatively. Stack-up is computed both worst-case and statistically, GD&T is applied per ISO 1101 or ASME Y14.5 with datum schemes that reflect how the part is actually located in the fixture, and tolerances are allocated against demonstrated process capability rather than distributed evenly. Specifying a tolerance the process cannot hold at a capability index of 1.33 is a decision to sort in production.
Design for Six Sigma, applied honestly, is the practice of writing down the transfer function between design parameters and the responses the customer cares about, and then choosing parameter values that put the response distribution safely inside specification given the variation the process will actually deliver.
Engagements run the identify-design-optimise-validate sequence: customer and regulatory requirements translated into measurable engineering characteristics with target and specification limits; concept selection against those characteristics; a transfer function built analytically, from simulation, or from designed experiments; then robust parameter design to find settings where the response is least sensitive to the noise factors that cannot be controlled — ambient temperature, material lot variation, installation workmanship. Monte Carlo propagation over the parameter distributions gives the predicted defect rate before a prototype is cut.
The output that matters is not a certificate. It is a design where the dominant sources of variation have been identified and either controlled or designed around, and where the residual risk is quantified rather than hoped about.
Design FMEA is done to the current AIAG-VDA seven-step structure, which is a genuine improvement on the older approach for one reason: Action Priority replaces the risk priority number. RPN arithmetic allowed a severity-10 failure mode to be ranked below a nuisance because occurrence and detection ratings were generous. Action Priority forces severity to carry weight it cannot be traded away from.
Two disciplines make the difference between a useful FMEA and a compliance document. First, structure and function analysis before failure analysis — failure modes are derived from the function each element performs at each interface, so the exercise systematically covers interfaces rather than components. Second, live linkage: design FMEA feeds process FMEA, process FMEA feeds the control plan, and a detection control claimed in the FMEA must exist in the control plan or the rating is fiction.
Reliability work goes further than a stated MTBF. Failure data is fitted to Weibull distributions and the shape parameter is read for what it says — below one indicates infant mortality traceable to workmanship or process control, above one indicates a wear-out mechanism that determines a service interval. Accelerated life testing is designed around the physical mechanism being accelerated: Arrhenius for thermally activated degradation, Coffin-Manson for thermal-cycling fatigue of joints, with the acceleration factor stated and the assumption that the mechanism has not changed at the elevated stress explicitly checked.
A test protocol should be written to find the failure, not to demonstrate the pass. The practice designs validation programmes that combine the mandated type tests for the sector — dielectric and partial discharge measurement for high-voltage apparatus, shock and vibration and environmental qualification for rolling stock equipment, ingress protection and salt-spray exposure for outdoor assemblies, abuse and propagation testing for battery systems — with margin discovery tests that deliberately go past specification to locate the design limit.
Knowing where the design breaks is more useful than knowing that it passed, because it tells you how much margin the next cost reduction can consume. Where step-stress or highly accelerated testing is appropriate, it is run early on prototypes, when a design change is still cheap.
Prototype work covers concept and detail design, supplier selection for prototype fabrication, instrumentation planning, acceptance criteria definition, test witnessing, and failure analysis with root cause established physically — metallurgical, fractographic or electrical — rather than inferred from a symptom.
Independent design review is most valuable at two moments: before tooling commitment, and after a field failure. In both cases the reviewer's usefulness comes from not having been in the room when the assumptions were made.
Reviews are conducted against a written checklist appropriate to the product class and produce findings ranked by consequence, each with a recommended action and an owner. Value engineering follows function analysis rather than component-by-component cost cutting: cost is allocated to the functions the product performs, functions carrying disproportionate cost relative to their contribution are identified, and alternatives are generated against the function rather than against the existing part. This is where the design practice and the practice's TRIZ work overlap, and function analysis carried out properly frequently exposes components that exist only to correct problems introduced by other components.
The practice's record includes 100+ patents delivered (including PCT filings) and 25+ simulation projects across nine verticals. NaraNova Tech LLP, incorporated May 2026, is based in Vadodara, Gujarat, registered under Startup India / DPIIT (DIPP271026) and Udyam (UDYAM-GJ-24-0239773), and GST registered.
This site uses only cookies that are strictly necessary for it to function. We set no analytics or marketing cookies unless you accept them. See our Cookie Policy and Privacy Policy.