Planning R&D Projects?

[pkelecy]

Hello All,

I work for a small engineering firm that does mostly R&D and prototyping projects. As the one responsible for developing project plans for these, I've really struggled with how best to do this. Some of the factors that I've found that make these hard to plan for include:

1. tasks often involve doing something that's never been done before (e.g. develop a new computer model). Since it's never been done before, the basis for estimating how long it will take is often poor (and, needless to say, it always seems to take longer than was estimated).

2. the sequence of tasks to perform often depends on the results obtained in one or more earlier tasks. A good example is a current project I'm planning for in which a contractor will be applying a coating to a key component in a device we're developing. If the coating works as expected, we'll use it. If not, we'll need to change the design of the component.

3. work efforts are iterative. For example, we may start with a "preliminary" design which is than modeled and analyzed (say, for stresses, temperatures, power output, etc). Based on those results, we may (and usually) need to go back and revised the design to correct problems. This *may* need happen several times.

Given all these uncertainties, does anyone have any recommendations (or know of any good references) that would help me get a better handle on this. Because we use our plans to estimate time (delivery) and cost to our customers, it's important that we estimate accurately. But as you can see, the natural of these projects makes that difficult. Most of the project management texts I've looked at assume (or seem to) that the work required is well define and sequential (nothing approaching my situation).

Thanks for any help. I appreciate it.

Pat

[Bernard]

I have always believed that projects based upon a creative process demanded a different paradigm for project management than the typical EPC method where plans involve discrete tasks that are easily estimated.

Have you looked into adapting any of the newer IT methodologies such as Agile, Extreme, etc.?

There is an old discussion on the Planning Planet regarding Engineering Design Vs Construction Projects, but I don't think it delves into issues faced with R&D projects.

I found this eCognition system while looking for something else. Perhaps it has some application for your case? Check out the Engineering Analysis and Planning/Scheduling papers.

[jjf_1975]

Hi Pat

I used to do a lot of scheduling of software development R&D projects - unfortunately it's part of the territory that the projects don't often run to plan because of so many unknowns.

A few things we did was...

1. split the project into distinctive phases of Concept Development, Planning, Implementation. We never tried to plan the entire project.
Concept Development was where the engineers did most of their background research. The planning phase, was not only to plan the Implmentation PHase but to do further research on those areas were they still had a high level of uncertainty...by the time we got to implmentation the engineers had reduce their level of uncertainty to a manageable level.

2. we built in contingency based on the phase of the project Concept 30%, Planning 20%, Implmentation 10%

3. use spiral planning methodology not a waterfall approach

4. use other metrics from similar project where possible, although the metrics are not for exactly the same work. For example, it takes X minutes to write a line of code in C++, we estimate this project will be 1000 lines of code, therefore multiple X minutes by 1000 lines of code.

There are also metrics regarding the number of requirements converts to so many use cases, each use case = x hours etc...

Here is a couple of website that may be of some assistance.
http://www.ispa-cost.org/PEIWeb/
http://www.softwaremetrics.com/Articles/estimating.htm

Regards
Joanne

PS: the experience software engineers have these kind of metric in their heads...it just hard to extract it out of them!!! It comes down to asking the right questions...and to do that you need to understanding software development process.

[pmkb]

martinig posted a link to another article on estimating with use case points.

This page defines different project lifecycle models (with prod and cons) including spiral planning and waterfall methods: http://www.business-esolutions.com/islm.htm