If you landed here directly, check out the first three posts of this series of - The complete industrial designer:
Stage 1: Research & observation
Stage 2: Sketching
Stage 3: Form & proportion
Stage 4: CMF
Somebody was once showing me a protruded block for an RO model we were working on more than a year ago. It looked good on screen, as they usually do, and I asked what happens if the width goes from 60mm to 65 or 75mm. He said give him a day. It took three, and the proportions came back wrong, because the thin groove around the border was too thick for my liking and I wanted it to be of different width for different block widths.
Some context. I don't model for a living and I never have. I run the products function, which means I review a lot of geometry, I sign off on things I couldn't build myself, and I've been wrong often enough to know which questions actually find problems. I've built simple things in Shapr3D, enough to have felt the difference between a model that rebuilds when you change a number.
So this isn't a modelling tutorial. It's what I've learned to ask for, and what happens when nobody asks. This, in turn, is how I would have wanted the industrial designers in the team to think about when modelling.
Where to model it
We run Rhino, and Rhino on its own is not parametric. There's History, and it records some operations, and it breaks on file transfer and on plenty of fairly ordinary edits. So two sketches have no relationship to rebuild from, which is exactly why trying a modification naturally takes days when it should have taken hours.
When a change takes three days, the instinct in a review is to push on the timeline, or judge the speed of the designer. Neither are true problems to solve for. It's a workflow problem.
Grasshopper is the parametric answer for Rhino, as I learned recently from someone on the team, and for this bottle it's probably better than it first looks. It's a graph rather than a feature tree, so the skeleton becomes sliders and the whole body can rebuild off those.
The cost, and I'm told this fairly consistently, is that a Grasshopper definition goes unreadable quite a bit faster than a feature tree does. A definition only one person can open is the same problem as a model only one person can change. So the thing I'd ask for are: labeled groups, sliders in one panel, and a written note of what each one does. That I can check without knowing how any of it works.
Then SolidWorks and Creo, which our the DE team uses and where the cap would get developed. A latch and a spring and a hinge want a feature tree and a proper assembly and draft analysis and mould tooling, and that's a fair description of what SolidWorks is for.
Before you start modelling
Write down the handful of dimensions the whole product gets built on like width, height, waist, neck, cap height, where the water line sits.
Give each one a plain name, and use the name everywhere.
Write down the relationships too like the groove is a tenth of the block width. The fillet is half the wall thickness, and so on.
Mark who owns each of the above numbers or specifics. ID owns or engineering owns, or is it shared. So when something changes, everybody knows whose call it was.
Agree on the same list, with the same names, across all tools before anyone starts.
How it gets built
Make simple placeholder shapes for everything that has to fit inside, before drawing the bottle. A cylinder the size of one litre, the air gap above it, the seals, and a block the size of the mechanism. Crude shapes are fine, they just have to be the right size and in the right place. Then draw the bottle around them.
If it's built in Grasshopper (in reference to what I saw the potential could be), I should be able to open it and understand the controls which should be labelled in plain words, all in one place, with a note saying what each one does.
The handoff, which is the part I own
Write down who owns what, and don't let anyone change someone else's without saying so be it overall shape, width, height and proportions when it comes to ID. Wall thicknesses, internal structure and anything to do with the mould are engineering's ownership.
The neck needs the most collaboration, because it belongs to both ID & DE. The thread, the seal and the latch all sit here, so a change on one side may break the other.
Testing the workflow
Ask for the width to go from 78mm to 80mm, and time how long it takes to come back. If it's in minutes, then it's ok. If it's a day more then the workflow should be re-evaluated.
The workflow above should make you more efficient and save time to do better things. Going beyond the basics like asking to make the groove thinner when the bottle goes wider in minutes is when we might know we've gotten the workflow optimised.
The surface
Think about the 'feel' of the bottle for a moment. When two curved surfaces meet, they can meet at different degrees of smoothness. They can just touch, which leaves a visible line, and that's fine when you meant to have an edge there. But the curve changes rate suddenly. Or they can genuinely flow into each other. One version may make people feel the product looks well made vs. it's just a bottle put together.
There's a second thing which is that a bad curve makes a bad surface. On a metal bottle a slightly rough joint looks a bit cheap and on a clear one it reads as a manufacturing fault, because you're seeing the front surface, the back surface through the material, and everything bending through the wall, all at once.
Some mould considerations
The main one is draft. Every moulded part has to come out of its mould, and if the sides are perfectly vertical they grip and won't release, or they release with scrape marks down the side. So the sides get a very slight taper. There's split mould option too but unless we figure out how to deal with the parting line on a clear body, let's stick to tapers for now.
Deliverables should also include
The list of key dimensions, current, with who owns each one.
A short change log. What moved, why, and what it broke. One or two lines per revision.
Version and date on every file and every drawing. So we always know which one we're looking at.
A table of the gaps and clearances, using the same names as the model.
A list of what's still undecided, so open questions are visible rather than quietly assumed.
A cut-through of the neck and cap at twice size, showing where the seal sits and how much it squashes, the thread, the latch, and a thumb drawn on the button so the scale is obvious.
A cut-through of the waist showing the wall thickness through the curve.
An exploded view of the cap with every part visible and counted.
Renders of full, half empty, and empty bottles. A clear bottle is three different-looking objects through the day.
Renders with the real colours applied, under daylight and under a warm indoor light, side by side.
Renders with a hand in shot, for size reference.
Render of the bottle placed next to Larq, Muji and Stanley at scaled proportions.