Machining Setup Estimating Time

Have you searched for anything, yet? Like on Search GlobalSpec, or even Google?

As a general rule, we don't do homework here. A little independent, creative thought won't hurt you.

empirical and practical experience. When I worked at the shipyard setup could take weeks (including design and fabrication of jigs and fixtures) for 4-6 hours of actual machine setup and macining of part in question

If you register on this site, you will have the ability to privately email me by clicking on my name above- it is a link.

After I get to know you and better understand your thesis, I may be able to help you on this.

milo

Milo and Guest -- You can count me in on this to the extent that I can help.

I have an original copy of Ostwald's AM cost estimator, 4th ed, and used its approach to estimating costs of machined parts for a good piece of a year at the end of my working career. It doesn't say too much about programming time and even if it did the constants for the time it was published (1980's) would be hopelessly dated given the progression of CAM software and machine control in the succeeding 20+ years as well as the transition to 3d solid modeling and the ability of CAM software to directly accept 3d engineering design files. This is an area I can't help with very much as I lack that kind of experience.

But the important thing here is that this rare book will be a useful reference for our guest in the work leading up to his thesis presentation.

It's been my impression that the information on this subject is mostly in the form of closely held data used to support expensive proprietary cost estimating software or service providers. I don't think a lot of useful info will be found on the internet. The people who sell these things may have something to say about the time needed to complete an estimate; but for the purposes of a scholarly dissertation such claims will likely be tarnished by the obvious bias of one selling an estimating product or service.

Hopefully one of our other CR-4 members with recent solid modeling and CNC programming experience will be able to join this effort and give us the benefit of his own experience.

The great unknown in estimating the programming time is to what degree the programmer must spend time interacting with the part designer over issues of manufacturability of the design as well as quality and integrity of the computer solid model. Different solid modeling software products may have widely varying characteristics in this area.

About Ostwald's approach to cost estimating; it was "newthink" for its time and a break from traditional cost accounting methods of an earlier time (like when I took the subject as an elective in my engineering undergrad program in the late 1950's). Fifty years ago the cost accountants were into materials, direct labor, fixed and variable overhead (burden as they called it) and burden variances. The result was "standard costs for each part of a product. At the other end the time and motion studies of manufacture of individual parts provided data observations to enable production engineers to improve processes as well as form the basis of standard times needed to support incentive programs popular in union contracts for mass production industries.

What Ostwald suggested was that another less exact approach was needed to provide data for a wide range of proactive management decisions for which cost estimates were good enough as opposed to exact "costs" of the accountant. Ostwald's method used an assigned cost per unit time for all labor and overhead for each unique machine center and could easily include profit margins to make estimating and quoting easy for a job shop. Then he said let's look at all the fabrication operations as well as the setup activities for that machine center and determine the actual run time for each unique operation as a function of some easily measurable factor such as part geometry. Trivial and second order effects were simply averaged in since this was for estimating and not accounting purposes.

Ostwald then put all of the various factors and how to use them for many processes into the AM Cost Estimator. The real basis of this approach is to simply break down all the factors in the cost of a manufactured component and sum them up for each separate process work center the part passes through.

Another important unknown relative to the subject at hand is the estimation of tool setting time and the manner in which it is apportioned to the cost of each distinct part. Our OP may not be seeking our help in this area but I think it is worth mentioning since the associated costs may be an important factor and costly variable in the management of manufacturing costs. Since tool holders carry a unique identification in modern CNC processes it is entirely possible to both estimate and assign costs to each individual and assign those costs on the basis known factors such as cutting time, depth of cut, feed rate, material, tool life and whatever else is quantifiable.

It is worth noting at this point that Ostwald's method is highly adaptable to spreadsheets for anyone with ability to construct macros.

Ed Weldon

Ed, This is an excellent response, And yet the polar oposite of MARKARS equally cogent response.

Guest, welcome to manufacturing.

Milo

agreed, manufacturing, it is so fluid nowadays

Well.....

My CIT Supervisor (Continuous Improvement Team) will say that however long the setup/change-over takes "We [read you] need to get better".

So picka a time, any time, and it will be too long a time.

"Guest" ...if that's your real name....today's your lucky day because I'm going to advance your understanding about 5 years for free.

You need time for CNC programming, loading and unloading time, so do about a million other shop owners, managers and supervisors.

Forget about books and the internet your going to have to humble yourself and ask an actual human being a question...to their face...and not interrupt them when they rant a little....take notes...buy them coffee....and say thank you when it's over if you want any meaningful information. I see this is for a thesis meaning no one's making or loosing money on your 1000 foot overview so this will not be as painful as real world because your job and a projects profit/loss is not on the line.

Every CNC program is different and is driven by what the scope of work is and what you have to work with. Here's a short list: CAD/CAM software...if any, CNC machines "type / quality / condition" cutting tools available, programmers experience, setup mans experience and the toys he has at his disposal, operators experience.

You need to talk to a shop Foreman and remember he's not laughing with you...but at you.

No disrespect intended but you can't put 20+ years of hands on experience and dozens of variables into a formula that will achieve anything close to what a seasoned veteran can shoot from the hip with in a few seconds.

"You need to talk to a shop Foreman and remember he's not laughing with you...but at you."

I'm thinking a lot of shop foremen laughed at Fredrick Winslow Taylor when he presented his ideas on Scientific Management over 100 years ago.

The subject at hand is an extension of Winslow's approach in one highly variable area of manufacturing, Production Engineering. It's important to note that any estimating process is almost by definition inaccurate since it takes place before the event. The real question is whether the estimate will be so inaccurate as to be of less value than the effort that goes into producing it. The very fact that CNC machining is a commercial success proves that its cost can be predicted with sufficient certainty to make the process of quoting a fixed price for machining services commercially feasible. And this in turn depends on a large percentage of machining jobs that the shop quotes fitting a predictable routine.

It appears to me that our guest is trying to quantify the routine tasks in programming a CNC machining job and relating them to easily measurable attributes of the prospective job. This has always been a necessity in management of large machine shops not only for cost control of individual jobs but also for shop scheduling in a competitive business where resources should not be so abundant that they stand idle much of the time.

In a small shop with just a few skilled and conscientious workers these approaches tend to be more trouble than they are worth given that communications are easy and close and the machinist usually programs his own work. This makes his judgment of time requirements better than any model can predict most of the time. It's a whole different situation in the large shop with many layers of management, a wide range of worker attributes and a fast paced and varying workload.

There are lots of easily observable attributes of any prospective job and specifically the programming and planning it entails that can be observed at the quotation stage and either quantified directly or placed into categories for which constants have already been developed from past work history. Some examples:

1. Format of the spec. (drawing, computer model, written spec, actual piece to be duplicated)

2. Size of the spec (sheets of drawings to be studied, word count to be read, number of dimensions on the drawing, etc.)

3. Quality of the spec (history of customer's engineering quality and responsiveness to questions from vendors, specific software used for computer models and local experience with how well it translates into CAM software, amount of debugging needed)

4. Start up time for CAM software and time metrics for various types of data entry

5. Time CAM software requires for each predictable activity in the programming process.

6. Time required for the programmer to deal with materials and outside secondary process requirements

7. Consultation time with other people in the shop to answer issues unknown to the programmer.

8. Uptime availability of programming software and hardware resources. (an issue if there are more programmers than computer stations or licensed software packages to run on them)

9. Additional production engineering activities to provide tooling or process development for a complex job.

10. Time to review results of first piece production and measurement and make needed corrections.

11. Time required for operator programming activities at the work center.

All of this stuff is subject to quantification based on historical observations and inclusion in an estimating spreadsheet. I'm sure there are lots of other factors I haven't thought of.

Ed Weldon

I don't have much to add, except that I remember from my manual machining days that the object was to take a piece of metal that was too big and make it smaller, and eventually end up with a "part", with recognizable features and some utility.

This was done by using a cutter to remove metal. The path the cutter took was controlled by the cranks. The cranks were controlled by the machinist. That "tool path" is now the basis of cnc machining. So, it comes down to programming the tool to make as few passes over the material as possible while still ending up with the desired configuration and a usable part.

I think you may be able to formulate some generalities, but I believe that experience is the only way to get close to actual programming time. Maybe a shop would let you review some of their historical files and you could get a rough idea of programming time that way.

You still can't copy everything off the net, yet, and I'm not convinced that copying information off the net aids an actual learning. Maybe learning isn't the point any more. Maybe it has just come down to information gathering (harvesting) as the goal of education.

Steelers by 3.

"Steelers by 3."

Bad news....and this data is from a FORMULA so it's got to be accurate enough to be the farm on....right....

ESPN NFL:

Sports books in Nevada made Green Bay a slight favorite in the Super Bowl over Pittsburgh on Sunday night based on the Packers' popularity among bettors during the past few weeks. The Glantz-Culver line made the Packers an early 2½-point favorite and put the total score at 45½ points for the Feb. 6 game in Arlington, Texas.

nothings a given