Machine Design #26: Estimating Design Cost and Engineering Hours — From WBS to Actual-Hour Control

When quoting machine design work, the hardest question is usually not "what is the hourly rate", but how many hours does this job actually need.

Many companies still estimate the design fee as a percentage of the machine price. That can work as a final sanity check, but it is not reliable enough to be the main basis. Two machines with a similar build cost can differ enormously in design hours: a machine reusing 80% of an existing structure is nothing like a new development that requires a site survey, several trial concepts and changes of bought-out equipment along the way.

To quote with less deviation, an engineer has to break the work into small enough items, write down the assumptions, include checking and revision time, and then compare the estimate with the actual hours after every project. That is how a durable quoting capability is built, instead of relying on the feeling that "if I did it, it would take about this long".

Do not take a fixed percentage of the machine price as the design fee

You will hear statements such as:

The design fee equals a certain percentage of the machine price.
Drawings account for a certain share of the total equipment value.
The more expensive the machine, the higher the design fee.

These ratios only mean something when the historical data set is large and the machine types are reasonably similar. In practice, design cost is driven by many factors that are not proportional to the build cost:

How novel the mechanism is.
The quality of the incoming specification and layout.
The number of units to be designed.
The number of mechanical, electrical, pneumatic and software interfaces.
How much previous design can be reused.
The number of Design Review (DR) rounds.
How often the customer changes requirements.
The availability of CAD data from equipment manufacturers.
Requirements for documentation, BOM, assembly drawings, detail drawings and handover files.
The capability of the designer and of the checker.

Machine number 1 and machine number 2 of the same product line should not carry the same design fee either. Machine 1 usually carries concept development, principle verification and the first round of troubleshooting. Machine 2 can reuse most of the data, but still generates revision hours based on feedback from machine 1.

So a percentage of the machine price should only be a cross-check, never the primary estimating method.

Separate three figures that often get mixed together

When quoting, keep three concepts clearly apart.

Engineering hours

This is the total time genuinely needed to complete the scope:

Collecting requirements.
Surveys and meetings.
Developing the concept.
Calculations and component selection.
3D modelling.
Producing drawings.
Checking.
Revising.
Coordination and document preparation.

Internal cost

Internal cost is not simply hours multiplied by a single rate. If the project involves several personnel levels, calculate by work group:

Internal labour cost
= Σ(Hours of each work group × Corresponding internal rate)

Then add the direct costs:

Internal cost
= Labour cost
+ Outsourcing cost
+ Travel cost
+ Any additional software or licence cost
+ Other direct costs

Customer price

The customer price has to cover internal cost, management time and commercial risk:

Customer price
= Internal cost
+ Project management cost
+ Risk contingency
+ Profit

Without separating these three layers, an engineer can estimate the hours fairly accurately and the company can still lose money by omitting management, outsourcing or unrecoverable change work.

Start from scope and assumptions, not from a number

Before filling in any hours, clarify at least the following:

Is this a completely new design or an improvement to an existing machine?
Does the customer have a complete specification?
Is there an overall layout and are the interface positions with other equipment defined?
Who is responsible for electrical, software, pneumatics and safety?
Is a site survey needed?
How many internal DR rounds and how many DR rounds with the customer?
What are the deliverables?
Are manufacturing drawings, BOM, pneumatic diagrams, assembly documentation or maintenance instructions required?
How many revision rounds are included in the scope?
Is CAD data available for the bought-out equipment?
Does the delivery date allow the proper sequence, or will it require more people working in parallel?

Where there is no answer yet, write it down as a quotation assumption. Assumptions are not an afterthought; they are the basis for deciding when a new request becomes additional work.

Break the work down with a WBS before estimating hours

A WBS — Work Breakdown Structure — divides the project into smaller items. For machine design, three levels usually work:

Whole machine
→ Unit
→ Sub-unit or specific work item

For example:

Assembly machine
├─ Part feeding
│  ├─ Part magazine
│  ├─ Separating mechanism
│  └─ Confirmation sensor
├─ Transfer
│  ├─ Conveyor
│  ├─ Stopper mechanism
│  └─ Locating mechanism
└─ Assembly station
   ├─ XYZ axes
   ├─ Working head
   └─ Product holding jig

The finer the breakdown, the less the hour allocation depends on feel. And when the estimate overruns or underruns, you can identify which item was wrong instead of only knowing that the project total was off.

Each group of hours has to be counted separately

Work before design starts

Many quotations begin at the moment CAD is opened, while the project has already consumed time before that:

Reading the requirements.
Checking the input data.
The kick-off meeting.
The site survey.
Taking dimensions.
Photographs and survey records.
Confirming interfaces with existing machines.
Preparing DR content.

These hours are part of the design work, even though no CAD window is open yet.

Concept design and overall design

This phase usually includes:

Defining the working principle.
Comparing alternatives.
Cycle time calculation.
Space checking.
Layout arrangement.
Defining the product flow.
Choosing the drive concept.
Safety risk assessment.
Producing the general arrangement or plan drawing.

A mistake at the overall stage propagates into every detail drawing. Do not cut concept design hours just to quote low and then make it up in later revisions.

Detail design per unit and sub-unit

Each unit should be broken down at least into:

Calculation and equipment selection.
Adjusting CAD data for bought-out equipment.
3D modelling.
Checking stroke and collision zones.
Detail design of machined parts.
Design of plates, frames or weldments.
Assembly drawings.
Detail drawings.
BOM.
Interfaces with neighbouring units.

A small unit with many interfaces can take more time than a large but self-contained structure.

Producing drawings

Drawing work is not just the act of generating a drawing from a model. One drawing may require:

Editing the model to suit the manufacturing process.
Choosing dimensional references.
Arranging views and sections.
Dimensioning functional features.
Specifying dimensional tolerances.
Specifying geometric tolerances.
Specifying surface texture.
Specifying material.
Specifying heat treatment.
Specifying surface treatment.
Checking threads, holes, chamfers and radii.
Checking manufacturability and inspectability.

A simple drawing can be finished quickly. A drawing with several datums, geometric tolerances or special treatments has to be assessed differently.

Drawing checking and design review

A plate with two holes can be drawn in a few minutes, but checking is not just confirming that two holes are present. The checker also has to confirm:

Do the holes match the bought-out equipment?
Is the mounting distance per the catalogue?
Can a tool reach in to tighten it?
Is there any conflict with a neighbouring unit?
Are tolerances sufficient for the function without being unnecessarily tight?
Are the material and treatment appropriate?
Does the BOM show the correct model and quantity?
Are the 3D data and the drawing consistent?

The checking ratio should not be fixed for every project. Use reference ratios by risk level:

Condition	Reference checking hours vs drawing hours
Repeat drawings, well standardized	10–15%
Normal design work	15–25%
New machine, many interfaces	25–35%
Precision equipment or high requirements	30–50%

These ratios are not a mandatory standard. Each company should calibrate them with its own data.

Revisions and changes

This is the group most likely to blow the estimate.

Revisions arise because:

The selected equipment cannot be purchased.
The lead time is too long.
Stock components have to be used.
The customer adds a function.
The overall layout changes.
Another unit changes its interface.
A DR outcome requires a structural change.
Testing shows stiffness or stroke has to change.

Distinguish between:

Fixing a design error: the designer's responsibility.
Adjustment within the agreed scope: counted within the included revision rounds.
A change of requirement or scope: has to be considered for an additional quotation.

If the quotation does not state the number of revision rounds, almost any change will be read as "already included in the price".

Meetings, coordination and work that is not on CAD

The following are commonly forgotten:

Internal meetings.
Customer meetings.
Supplier discussions.
Preparing DR material.
Building the schedule.
Preparing the quotation.
Data management.
Exporting files and checking formats.
Preparing handover documentation.
Answering questions from manufacturing and assembly.

Do not call these "non-design hours". They do not create the model directly, but they are what turns the model and drawings into an engineering deliverable that can actually be built.

Estimating detail drawings

For converting 3D models into 2D drawings, break it down as:

Detail drawing hours
= Model repair or clean-up hours
+ Drawing creation hours
+ Checking hours
+ Other handling hours

Factors that increase the hours:

The source model has errors or features that are hard to edit.
There is no clear dimensional reference.
Many sections are needed.
Many geometric tolerances are required.
Heat treatment or surface treatment is involved.
Weldments or sheet metal need their own notation.
The drawing has to follow a customer-specific format.

Do not apply a fixed time to every drawing based only on part size.

Estimating part tracing work

When rebuilding a part from a PDF or a 2D drawing:

Part tracing hours
= Reading and verifying data
+ 3D modelling
+ Drawing creation
+ Checking
+ Clarifying missing information

Source quality has an enormous effect:

Source quality	Characteristics	Effect
Good	Vector PDF, complete dimensions and tolerances	Can be modelled directly
Medium	Clear drawing but some information missing	Requires inference and confirmation
Poor	Blurred scan, contradictory dimensions	Sharply increases checking and communication time

If a drawing is missing dimensions, the engineer should not guess and then treat that as official data. Unclear points have to be recorded as a question list or as assumptions.

Estimating bought-out components

Time spent handling bought-out components is design time too.

Common situations:

The manufacturer provides 3D data.
Only 2D drawings exist and 3D has to be modelled.
3D data exists but needs conversion, clean-up and simplification.
A request has to be sent to the manufacturer.
No reliable data exists and a simplified model has to be built from catalogue dimensions.

The work can include:

Finding the correct model.
Checking the revision.
Downloading the file.
Checking the units.
Checking the coordinate system.
Converting STEP, Parasolid or an intermediate format.
Removing unnecessary internal detail.
Checking mounting dimensions.
Checking stroke and restricted zones.
Recording the data source.

Separate labour hours from waiting time. Waiting three days for a manufacturer's reply is not 24 working hours, but it still affects the project schedule.

An overall formula for estimated hours

A simple structure works:

Base hours
= Preparation hours
+ Design hours
+ Drawing hours
+ Checking hours
+ Revision hours
+ Meeting and coordination hours
+ Data preparation hours

Then apply a risk factor:

Estimated hours
= Base hours × Risk factor

The risk factor has to reflect uncertainty, not be an arbitrary mark-up.

Project type	Reference factor
Repeat of a stable design	1.05–1.10
Improvement of an existing machine	1.10–1.20
New machine, requirements reasonably clear	1.20–1.30
New development, many undefined points	1.30–1.50

If the specification is still full of gaps but the factor stays low, the estimate only creates an appearance of precision without actually controlling risk.

Converting hours into engineering days

Do not assume that an eight-hour working day yields eight effective design hours. Real time is also consumed by:

Company meetings.
Email and correspondence.
Supporting other projects.
Data management.
Administrative work.
Interruptions during the day.

The formula should reference a clearly defined value:

Engineering days
= Estimated hours / Effective hours per day

For example, if the company uses 7 effective hours per day, use that consistently throughout the file. Do not calculate the top section at 8 hours while a note underneath says 10 hours per day.

A worked example

Suppose a small machine unit has this base estimate:

Work group	Hours
Preparation and survey	10
Design and equipment selection	80
Drawing production	56
Checking	14
Expected revisions	16
Meetings and coordination	12
Data preparation	8
Total base hours	196

It is a new design but with a reasonably clear specification, so a factor of 1.20 is chosen:

Estimated hours = 196 × 1.20 = 235.2 hours

At 7 effective hours per day:

Engineering days = 235.2 / 7 ≈ 33.6 days

These are engineering days, not calendar time. With two people working in parallel, the project schedule does not automatically halve, because it still depends on the precedence relationships between items and on checking time.

The figures above illustrate the method; they are not a universal norm.

Manage change from the quotation onwards

The quotation should state clearly what is included:

The number of internal DR rounds.
The number of DR rounds with the customer.
The number of revisions.
The type of data handed over.
The CAD format.
Whether a BOM is included.
Whether pneumatic drawings are included.
Whether manufacturing and assembly support is included.
Whether post-commissioning updates are included.

It should also state the cases that require re-assessment:

A specification change.
A layout or interface change.
A component change requested by the customer.
Additional units or functions.
A change of drawing standard.
A schedule change requiring more people or overtime.
Input data that does not match the original assumptions.

A detailed estimate sheet is not only there to justify the price. It is also the evidence of what was in scope and what arose afterwards.

Compare estimates with actual hours

An estimate only improves when it is verified.

After every project, record:

Estimated hours.
Actual hours.
The deviation.
The cause.
The improvement action.

The basic formulas:

Hour deviation = Actual hours - Estimated hours

Deviation rate
= (Actual hours - Estimated hours) / Estimated hours × 100%

Do not simply conclude that "the engineer was slow". Classify the causes:

The estimate missed work items.
The specification changed.
CAD data was unavailable.
The assigned skill level did not fit.
Checking time was underestimated.
Revision rounds exceeded the assumption.
The project was interrupted.
Manufacturing support was not itemized separately.

After a number of projects, the company will have real data for each machine family, each drawing type and each engineer level. At that point the risk factors and norms become far more trustworthy than a template borrowed from outside.

Common mistakes when estimating design hours

Counting only 3D modelling and 2D drawing time.
Omitting surveys, meetings and DR preparation.
Omitting time spent finding component data.
Using a checking ratio that is too low.
Not separating error correction from requirement changes.
Not limiting the number of revision rounds.
Using the same risk factor for every project.
Typing total hours by hand instead of using formulas.
Not agreeing on effective hours per day.
Mixing travel cost into engineering hours.
Not listing the exclusions.
Never comparing the estimate with the actual.

Checklist before sending the quotation

Scope

New design versus improvement is clearly defined.
The list of units and sub-units exists.
Deliverables are defined.
CAD and drawing formats are stated.
The responsibilities of each party are defined.

Input data

The specification is sufficient for estimating.
Layout and interfaces are confirmed.
Bought-out equipment data has been checked.
Unclear points are recorded as assumptions.

Hours

Survey and meeting hours are included.
Design and equipment selection are included.
Drawing hours are included.
Checking hours are included.
Expected revision hours are included.
Research and data preparation hours are included.
A justified risk factor has been applied.

Commercial and change

The number of DR rounds is stated.
The number of included revisions is stated.
The conditions for an additional quotation are stated.
Direct costs are itemized separately.
Project management and profit are included.
The validity period of the quotation is stated.

Conclusion

Estimating machine design cost is not guessing a number quickly. It is modelling the work before the project begins.

A good estimate has to answer:

What will be done?
Who will do it?
How many hours will it take?
What conditions are the assumptions based on?
How many checking and revision rounds?
When does something fall outside the scope?
After the project, where was the estimate wrong?

Do not try to make the number look accurate to the minute while the scope is still vague. What matters more is breaking the work down correctly, writing the assumptions clearly and maintaining the loop that compares estimates with actuals. Once enough data accumulates, quoting becomes an engineering capability of the organization rather than something that depends on one person's experience.

View all MINATA technical articles