Tag: Primavera

One of the biggest critiques/limitations in using P6 data is the lack of time phased resource assignment data. The only effective way to pull time phased resource assignment data into PowerBI (or even excel ) is to copy-paste from P6 into Excel. This is what I have recommended in the past and still what I would recommend for anyone moving forward. However, that does not mean that PowerBI can’t produce time phased data using a start date, end date, and profile. What follows is a simply guide on how I have tackled the problem (and the limitations I have run into)

Note: In all likelihood this problem has already been solved my many people in many different ways. I do not want to suggest this is “the way” to solve this. More so, I want to simply raise awareness of at least my approach and welcome comments and feedback on how to really solve the problem

Part 1 The Problem

P6 data does not contain time phased distributions. Instead, the backend (and inside XER) data only contains information about the activity, the resource, and the profile applied. So, our problem is to try to extract these data elements and generate a time phased distribution of the resource according to the profile and activity start and end dates.

I’ve taken a few liberties with the data above to try to focus on the:

Key Problem: how to allocate the resources assigned to an activity according to a resource profile?

I’ve seen this done in excel a lot although, i’ve never been comfortable with the excel solutions. Typically we count the weeks and distribute the hours equally to all the weeks. I know more complex files exist that allow for spreads using profiles. So to add to the problem isn’t nessessarily to simply spread per the profile, but to perform the operation inside PowerBI (or perhaps more to the point, inside Power Query).

At this point, I’d again love to call upon anyone who has a nice solution to include links about how you tackled this as what follows is just my initial stab at this.

Part 2: Getting the data into PowerBI

Firstly, although the data above is from a P6 XER file, I wanted to make this a bit more general. Therefore, I have created a toy model approach. Thus, here is what my source data looks like

Resource Profile Data:

We can run this through PowerQuery and convert it to a usable table. Note in the above I have created a PeriodCum field. This will be used to calculate an end date for each of the 10 periods required.

Activity and Resource Data:

In this example, I am combining the Activity and Resource data into just 1 table. Obviously if you were doing this formally, you would need to build a scheme to link the Resource Assignment data into the Activity level data

And running the above through Power Query we end up with something as seen below. Note I have added a calculation for the duration (in cal day) and have converted the date formats to numbers. This make the subsequent steps a little easier

Note: a critical hard step (for me at least) came duration this stage. Because we are breaking the duration into 10 periods and will ultimately be allocating a qnty per day to each each, if we have a fraction of a day (example a duration of 15 days), my method bombed. This caused an overlap of qnty allocation on the day of the overlap. As such, I have rounded the duration to the closest 10s.

Part 3: Time Phasing (where the magic happens)

The first step of generating the time phasing is to now split the activities into the 10 periods per the resource allocation. We do this by first merging tbl_activies with tbl_profiles using the ProfileDesc field. After expanding the result, we will end up with 10 records for each activity (corresponding to each of the 10 periods).

We will now want to calculate a start and finish date unique for each of the 10 periods. In the profile table is a PeriodCum field that we can use to multiply by the duration and then add that to the start date to get a finish for each period. The result table will now look something like this. Note, at this step, its good to now use the profile allocation for each period and multiple that by the hours_total field. This will give us a hours per each period. The last step will be the divide that by the period duration to calculate what will now be an hours per day for that period.

Lastly, we now want to perform 1 additional expansion to get the DAYS for each period. Here is the code I have used. This is a nice little bit of code that can generate a sequential list from a start to an end (we are using days, but its works for any beginning to end sequence)

= Table.AddColumn(#”Changed Type2″, “Custom”, each { Number.From([PeriodStart])..Number.From([PeriodFinish]) })

In our resulting table (see above), we can clearly see that the hours per day adjusts for each period based on the profile.

Part 4: Putting It All Together

We have our activity data, we have our time phased resource data, the last step is to generate a DIM_Date table that can be used to bin the days to either weeks or months (or quarters or years or any custom grouping defined inside the DIM_Date table)

I don’t want to get to into the DIM_Date table, effectively all we need is each day assigned to a Month-Yr for the purpose of generating a nice little graph below, which is ultimately what we are after.

Conclusion

Again, I do not recommend using this approach. Instead I am more interested in how others have decided to approach this. Personally, as I noted at the beginning, my recommendation is to use a copy-paste from the P6 Resource Assignment tab. Although, this time phasing approach can be used for other (non P6) applications. But alas, I believe there are much smarter ways to achieve the spread using the DIM_Date table and perhaps custom functions. In my research for my method, I ideally wanted a “working day” spread as opposed to the calendar day spread. There are some awesome custom functions that can provide an integer for the number of working days between 2 dates. However, even when taking that approach, I ran into additional complications in getting everything to work.

So, really keen for feedback!

thanks