Manufacturing and Scaling

facebooktwittergoogle_plusredditpinterestlinkedintumblrmailfacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

Some thoughts on how WIKISPEED works within or around traditional manufcaturing constraints:
Labor Pool: We invest thought each week into reducing the gap between anyone on the street and a productive Team WIKISPEED team member. We have re-factored our on-boarding process down to a 5 minute safety, tools, and process orientation after which people are able contribute to team where it is highest priority now and with a quality bar.

Here is an orientation video and shop tour at our Lynnwood, WA, prototyping and test shop:  http://www.youtube.com/watch?v=U_RxF3uEfWk.

Note, the tour goes through the manufacturing and prototyping flow in reverse order, and it also may be fun to note the manufacturing flow starts in 3 bays (2, 3, 4) simultaneously.

Automation and Flow: Our current inertia is compacting our operations into a floor space the size of a 40’ shipping container. The gaol there is to ship micro-factories to the areas where purchasing is happening now, and load-balance them in real time by semi truck or boat, similar to Microsoft, Google, Twitter, and Facebook server farms. The shipping container factory flow-plan currently requires a pair of two people with a maximum crew of 8 people. Currently the prototype-flow starts with setting a stack of aluminum extrusions on labeled templates on the CNC router, we then hit “go” and then go off to assemble wiring harnesses and engine components. Then we come back and the entire chassis and suspension has been kitted in one setup operation. This is similar to some of the laser operations currently built at an agricultural machinery factory I recently toured. The chassis is then self jigging with 32 bolts- we could automate that later with an industrial robot but currently it takes a pair about 10 minutes for a practiced pair. The TIG welder has an option for CNC control in our same CNC gantry, and we may explore that in the future; currently we have a 6 hour manual weld step there. meanwhile the CNC router, the same one, is now “printing” (subtractive rapid prototyping) our foam slabs for whatever shape this car body will be. By the time the welding is done the foam buck is ready for carbon to be laid on, which takes 8 hours. While the carbon cures we connect the engine wiring to the engine module and unit-test the engine module, and assemble the interior. The carbon cures 8 hours later and we trim the flashing, install the LED lighting and wiper, install the windows, and attach it to the chassis. Although we aren’t currently in volume scenarios, we are building our 5th and 6th cars right now in 2 hour chunks on Thursday nights, it looks like from start to finish we’ll be 3 8 hour shifts and we can have 3 cars in progress at a time per shipping container micro-factory. Automation does appear to be able to reduce the time, and Open Source Ecology is working on a low-cost owner-maintainable industrial robot (http://opensourceecology.org/wiki/Industrial_Robot). We are talking with them about prototyping it using what we’ve learned about Arduino control of linear actuators, and that could be awesome.

simplification and mistake proofing: Scrum methodology has a kaizen event/continuous improvement event called a Retrospective every 7 days. Every 7 days we collaborate as a team to create backlog items that simplify, reduce waste, decrease our environmental impact, reduce cost, increase ergonomics, increase self-explanation, etc. These are then put in the backlog with the same tasks as build and deliveries according to their priority. The time spent on implementing these tasks is not predetermined but is instead dictated by priority relative to the other work in the backlog. In addition to the weekly retrospectives, we hold retrospectives after key events like customer deliveries, test drives of new prototypes, auto-shows, etc. This is augmented by Test Driven Manufacturing, where tests fixtures are built for a new part before the part is even designed. By designing parts to pass tests, we avoid many types of complexity and have simplification up-front. This also lets us reduce any complexity that is not required to pass each test. Each test is from a customer’s perspective, something that is actually meaningful to them, such as fuel efficiency or how they will activate the turn signals.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>