# Make duration dependant or update it based on a cell

Starting with my first ms-project:

I want to plan the production of our machines at work. I only know the time it takes to produce one part (example: reference #1 takes 20 seconds). Now, I can calculate how many days it will take to produce 55.000 parts and input that in "duration" (which will be ~12 days). Can I upadte the duration manually depending on production? Something like this:

Starting date: 1/1/2010 Total parts needed: 55.000 Cycle time: 20'' Initial estimated production time: 12 days Finish date: 12/1/2010.

• Delay of production due to mechanical problems -

Now, finish date should be moved one day because instead of 12.000 I have produced 8000.

I know I can do this using days, but the only information I have from machines is how many parts we have already produced.

Can I do the calculations in different cells and update duration accordingly?

What will happen if a resource had to be used on the 12-1-10 and instead, it's still not available due to the delay? Will it move it further?

Thank you very much and apologies for the noob questions.

I think you are asking how to forecast remaining work and a new finish date based on progress to date. Should you let the schedule predict a new finish date based on the actual data you load? Should you use a formula such as with Earned Schedule or Earned Value or another method to predict a new finish -date? Should you estimate remaining work using both a top-down, bottom-up estimates?

There are multiple forecasting models you can use and each of those has varying degrees of reliability and validity making them all useful while also producing inaccurate results to some degree. Over time doing the same work, you might notice a forecasting model that is producing more accurate results than the others and you can begin putting more weight on that particular model's results.

For example, you predicted that you need to increase your duration by one day due to finishing 8,000 versus a planned 12,000. I am not sure how you arrived at one day; however, using Earned Schedule analysis on the data you provided, I arrive at the following:

Earned Schedule = 1.75 days (this means that instead of earning 3 days of work you only earned 1.75 days of work based on producing 8,000)

SPIt = 58% (this means you are operating at 58% efficiency based on plan)

iEACt = 21 days (this means that, assuming your 58% efficiency prevails, you will finish 55,000 units on day 21 instead of day 12, or 13 using your prediction).

So between these two models, your forecast ranges from being one day to as much as eight days late. Using more models will help you visualize your cone of uncertainty and then allow you to predict where you will finish with more credibility.

Taking another reading at day 5, you may learn your performance has improved, stayed the same, or further degraded and then you can use those models again to reforecast when you are finishing those 55,000 things.

• Alright. I know I have produced 8000 instead of 12000 because machine didn't produce for X amount of time due to a technical problem. However, I expect it to carry on working as planned from the beginning. From your explanation I understand that there are different ways. Will try to look for the one that suits my needs. Say I produce 1000 parts/day and I want to produce 10000 parts. After 3 days I will have produced 3000. On the 4th day, machine doesn't work. After the 5th day, I have produced 4000. My aim is to write "4000" on the 5th day and have duration updated to 11 days. – сами J.D. Oct 22 '17 at 16:46
• You have the right idea. Sounds like you're trying to do a bottom up estimate on the remaining work. – David Espina Oct 22 '17 at 17:35