Do you have a contractual requirement for using EVMS? Are you required to comply with ANSI 748? 

Stephen, there is now contractual requirement.  It is for an private owner organization with no requirement to comply to EIA748. 

Do you have a contractual requirement for using EVMS? Are you required to comply with ANSI 748? 

That's good, you have the freedom to re-plan each activity that was included in the replan approval even if it is behind schedule. 

Stephen, there is now contractual requirement.  It is for an private owner organization with no requirement to comply to EIA748. 

Do you have a contractual requirement for using EVMS? Are you required to comply with ANSI 748? 

Michael:

You raise several issues, so let me take them individually.

1. Re-Baselining - whether you rebaseline or not, the EVM curves will show a jump where the plan changes.  I believe it is necessary to rebaseline when the plan becomes significantly different or the schedule will no longer be reliable.  And when they rebaseline, everything that is still in the as-planned stage will get reworked, that's normal.  What I don't believe is normal is for a contractor to try to rework the full schedule starting from NTP - actualized data cannot be changed.
   1. A rebaseline, by its nature, will formally change the schedule, that's to be expected.
2. SPI returns to 1.0 and SV returns to 0 at project completion.
   1. This is part of the downside to running EVM on fixed price jobs where the planned value is the sum or the price to the owner, and not the costs of the contractor.  When EVM (C/SCSC - Cost and Schedule Control Systems Criteria) was originally envisioned, it was only used for cost reimbursable projects and true job costs were collected as Actual Costs.  But over the years, people realized that EVM is useful for invoicing so owner costs were loaded, and the BAC can never exceed the PV (change management excepted).  This means that in the last 30% or so of the project, the time metrics will improve regardless of the status of the project, even with very late predictions of completion.   
   2. You should check out Walter Lipke, who invented "Earned Schedule" which computes how late the project is by comparing where the planned value curve crosses the data date compared to the earned value curve.  This is similar to how we looked at it in the old graphical EV days, and works quite well, even when the project is trending late.
3. Taking on Earned Value
   1. Be aware that EVM run at the project level is just not all that useful because an early delivery of a large piece of equipment can make the metrics look good even when the project is in trouble.  You still need critical and near-critical path analysis because those are the activities that will cause immediate delay or experience a mid-period critical path shift and cause delay.  EVM is good for trending analysis, similar to duration overruns, missed starts, float erosion, and other methods that look at performance but not necessarily critical path delay.
   2. But when running EVM on filtered data sets like trades, submittals, floors, processes, critical paths, it can very useful and provide good insight not easily gained in other ways.  You can run EVM on anything that has activity codes to isolate the type of work
   3. With fixed price contracting, the time metrics are more useful than the cost metrics (even though time is presented through cost).  The better you do with loading costs, the more reliable you will find EVM metrics.  I like to see large equipment packages, like service equipment, boilers, light fixture packages, loaded into one-day delivery activities. That way when the equipment hits the job, it can be actualized and added to the invoice automatically.
   4. There are several RPs that you will find useful and should take a look at, RP82R-13, 79R-13, 80R-13, to name a few.  Also the Library is chock full of good papers on EVMS.
4. We use a technique that I call "duration-day" earned value (same name used by Murray Woolf, the only other person I know who has run something similar).  One of my colleagues, who has done a lot of EVMS aligned with EIA-748 global standards, really didn't like my approach, so I challenged him to co-author a paper and we did so this past summer.  Our conclusion?  That DD EV is very useful and can be implemented successfully.  I like it because our schedulers can do it with global changes and we don't need contractors to put in any additional work. 
   1. I've attached our paper for your use.

I don't know if this answers your questions, and you might want to post it in the EVM Community to ensure everyone who is engaged with earned value monitoring sees it.  

------------------------------
Christopher (Chris) Carson CEP DRMP PSP FAACE
Director of Program & Project Controls, Vice President
ARCADIS
Virginia Beach VA
(757) 342-5524
Chris.Carson@Arcadis.com
------------------------------

Michael:

You raise several issues, so let me take them individually.

1. Re-Baselining - whether you rebaseline or not, the EVM curves will show a jump where the plan changes.  I believe it is necessary to rebaseline when the plan becomes significantly different or the schedule will no longer be reliable.  And when they rebaseline, everything that is still in the as-planned stage will get reworked, that's normal.  What I don't believe is normal is for a contractor to try to rework the full schedule starting from NTP - actualized data cannot be changed.
   1. A rebaseline, by its nature, will formally change the schedule, that's to be expected.
2. SPI returns to 1.0 and SV returns to 0 at project completion.
   1. This is part of the downside to running EVM on fixed price jobs where the planned value is the sum or the price to the owner, and not the costs of the contractor.  When EVM (C/SCSC - Cost and Schedule Control Systems Criteria) was originally envisioned, it was only used for cost reimbursable projects and true job costs were collected as Actual Costs.  But over the years, people realized that EVM is useful for invoicing so owner costs were loaded, and the BAC can never exceed the PV (change management excepted).  This means that in the last 30% or so of the project, the time metrics will improve regardless of the status of the project, even with very late predictions of completion.   
   2. You should check out Walter Lipke, who invented "Earned Schedule" which computes how late the project is by comparing where the planned value curve crosses the data date compared to the earned value curve.  This is similar to how we looked at it in the old graphical EV days, and works quite well, even when the project is trending late.
3. Taking on Earned Value
   1. Be aware that EVM run at the project level is just not all that useful because an early delivery of a large piece of equipment can make the metrics look good even when the project is in trouble.  You still need critical and near-critical path analysis because those are the activities that will cause immediate delay or experience a mid-period critical path shift and cause delay.  EVM is good for trending analysis, similar to duration overruns, missed starts, float erosion, and other methods that look at performance but not necessarily critical path delay.
   2. But when running EVM on filtered data sets like trades, submittals, floors, processes, critical paths, it can very useful and provide good insight not easily gained in other ways.  You can run EVM on anything that has activity codes to isolate the type of work
   3. With fixed price contracting, the time metrics are more useful than the cost metrics (even though time is presented through cost).  The better you do with loading costs, the more reliable you will find EVM metrics.  I like to see large equipment packages, like service equipment, boilers, light fixture packages, loaded into one-day delivery activities. That way when the equipment hits the job, it can be actualized and added to the invoice automatically.
   4. There are several RPs that you will find useful and should take a look at, RP82R-13, 79R-13, 80R-13, to name a few.  Also the Library is chock full of good papers on EVMS.
4. We use a technique that I call "duration-day" earned value (same name used by Murray Woolf, the only other person I know who has run something similar).  One of my colleagues, who has done a lot of EVMS aligned with EIA-748 global standards, really didn't like my approach, so I challenged him to co-author a paper and we did so this past summer.  Our conclusion?  That DD EV is very useful and can be implemented successfully.  I like it because our schedulers can do it with global changes and we don't need contractors to put in any additional work. 
   1. I've attached our paper for your use.

I don't know if this answers your questions, and you might want to post it in the EVM Community to ensure everyone who is engaged with earned value monitoring sees it.  

Chris, you are saying that the EVM curves will show a jump where the plan changes.  From the various posts it seems it is allowed to amend the plan once it is approved and only rebaseline when there is some major change.  Basically for any approved change to the cost and schedule plan, it is okay to update the plan to reflect the new time phase plan for remaining work.  Maybe the term re-plan and baseline are different.  The second question is how is EVM used to monitor float erosion?  Not sure that is clear. EVM doesn't reference the schedule and only deals with cost.

Michael:

You raise several issues, so let me take them individually.

1. Re-Baselining - whether you rebaseline or not, the EVM curves will show a jump where the plan changes.  I believe it is necessary to rebaseline when the plan becomes significantly different or the schedule will no longer be reliable.  And when they rebaseline, everything that is still in the as-planned stage will get reworked, that's normal.  What I don't believe is normal is for a contractor to try to rework the full schedule starting from NTP - actualized data cannot be changed.
   1. A rebaseline, by its nature, will formally change the schedule, that's to be expected.
2. SPI returns to 1.0 and SV returns to 0 at project completion.
   1. This is part of the downside to running EVM on fixed price jobs where the planned value is the sum or the price to the owner, and not the costs of the contractor.  When EVM (C/SCSC - Cost and Schedule Control Systems Criteria) was originally envisioned, it was only used for cost reimbursable projects and true job costs were collected as Actual Costs.  But over the years, people realized that EVM is useful for invoicing so owner costs were loaded, and the BAC can never exceed the PV (change management excepted).  This means that in the last 30% or so of the project, the time metrics will improve regardless of the status of the project, even with very late predictions of completion.   
   2. You should check out Walter Lipke, who invented "Earned Schedule" which computes how late the project is by comparing where the planned value curve crosses the data date compared to the earned value curve.  This is similar to how we looked at it in the old graphical EV days, and works quite well, even when the project is trending late.
3. Taking on Earned Value
   1. Be aware that EVM run at the project level is just not all that useful because an early delivery of a large piece of equipment can make the metrics look good even when the project is in trouble.  You still need critical and near-critical path analysis because those are the activities that will cause immediate delay or experience a mid-period critical path shift and cause delay.  EVM is good for trending analysis, similar to duration overruns, missed starts, float erosion, and other methods that look at performance but not necessarily critical path delay.
   2. But when running EVM on filtered data sets like trades, submittals, floors, processes, critical paths, it can very useful and provide good insight not easily gained in other ways.  You can run EVM on anything that has activity codes to isolate the type of work
   3. With fixed price contracting, the time metrics are more useful than the cost metrics (even though time is presented through cost).  The better you do with loading costs, the more reliable you will find EVM metrics.  I like to see large equipment packages, like service equipment, boilers, light fixture packages, loaded into one-day delivery activities. That way when the equipment hits the job, it can be actualized and added to the invoice automatically.
   4. There are several RPs that you will find useful and should take a look at, RP82R-13, 79R-13, 80R-13, to name a few.  Also the Library is chock full of good papers on EVMS.
4. We use a technique that I call "duration-day" earned value (same name used by Murray Woolf, the only other person I know who has run something similar).  One of my colleagues, who has done a lot of EVMS aligned with EIA-748 global standards, really didn't like my approach, so I challenged him to co-author a paper and we did so this past summer.  Our conclusion?  That DD EV is very useful and can be implemented successfully.  I like it because our schedulers can do it with global changes and we don't need contractors to put in any additional work. 
   1. I've attached our paper for your use.

I don't know if this answers your questions, and you might want to post it in the EVM Community to ensure everyone who is engaged with earned value monitoring sees it.  

Michael:

Something that I thought I said but I don't think was clear, there is a large difference between running EVMS on fixed price projects compared to cost reimbursable projects, and I don't think that industry literature generally addresses fixed price contracts.  I suggested that you post in the EVMS community but remember that those folks are mostly the federal EVMS people, so they are fully attuned to following the EIA-748 standards for cost reimbursable projects where they are loading contractor's actual costs, which means that they rely on cost metrics as much as schedule metrics, whereas in the fixed price world, we tend to rely more on schedule metrics and recognize that there is value in the cost metrics but since those are simply percentages of the contract amount, they are most useful when the project is on time and budget.  Also, AACE and commercial/non-governmental best practices note that contingency is assigned to potential risks and is expected to be spent, and is visible to the project, whereas management reserves are hidden from the project and only expected to be spent for scope changes.  In the federal government world, that is reversed.  Just so you'll be aware.

I said that you can see a jump in the EVM curves if you do not re-baseline, so I think it is obvious and can be easily explained if you don't re-baseline.  In the fixed price world, from my perspective, I use the baseline and re-baseline cost loaded schedules only as a vehicle to hold the contract - meaning that it is my reference to contractually mandated cost and time.  So, if I have a change order that increases costs and/or provides an EoT, I would add an activity (or multiple if they affect different contractual milestones) that would carry the additional cost and time, and just link that at the end of the project.  That way when I reference back to the contractual re-baseline it gives me the right cost and time.  My most reliable comparison benchmark is the latest updated schedule as long as it is being reviewed and approved because that is the contemporaneous understanding of the critical path at each update.

But I believe one of the other large differences in running EIA-748 compliant EVMS for federal governmental projects is that the baseline schedule is of vital importance acting as the benchmark against which project performance is measured, so rebaselining is a harder task than in a fixed price non-federal project where you could just adjust logic in an update and compare it to the previous update.  In the fixed price world, I do not like changing logic in the baseline schedule to create a rebaselined schedule because the conditions are very different at the rebaseline point than at the original baseline.  I think it is dangerous to start looking at how the project would be planned differently in the baseline (at the time of creation of the baseline) because I think it leads to someone trying to determine "what the schedule would have looked like if it had been created to include these two years later changes".  I think that is what is dangerous because the field conditions are not the same and trying to recreate a new baseline that somehow is rooted in the original plan but now contains new conditions can be misleading.  I think it's better to keep the schedule revised to match the current plan and just compare the contract time and cost, while benchmarking against the previously approved "contract" schedule.  

But I think in the cost reimbursable world, (in my experience only) we tend only to rebaseline when there is a significant change to phasing or sequencing, or a significant change order, not for every change.  My perception is that most practitioners in that world keep the cost rebaselined but are not so fixed on the time.  Others will have more insight.

So, I think for practical purposes replan and rebaseline are reasonably the same for larger needs.  You could say that the contractor replans every time he updates the schedule and performs maintenance to keep the schedule serving as a good model of the work.  But I don't think most people rebaseline for every small change, but as I said, for me, the rebaseline is a current benchmark of the contract so I do rebaseline for every change order that has time or cost.

With respect to float erosion, remember that we have to monitor 1, the critical path since those are the activities that will immediately cause delay and prolong the completion predictions, 2, the near-critical path since those are the activities that might suffer a mid-period critical path shift, sneak up on the team (as they are focused more on critical paths) and cause a delay that also prolongs completion predictions.  But all the work must be installed at a rate commensurate with completing on time, and the way to monitor that best is with the various trending analysis methods.  Those methods include float erosion, missed starts and finishes, duration overruns, resource analysis, and earned value.  So, while EVMS doesn't measure float erosion (in fact, if you don't take into account the total float values, your EVM metrics will not be so meaningful.  The way that I merge the two needs is by running EVM on subsets of the schedule, such as the critical path, the near-critical path, the trades (to ensure that a trade not on the critical path is not allowing momentum to falter because he has lots of time before going critical), the various stages and phases, areas, and anything else with an activity code that allows filtering.  You might be surprised at how much more insight you get from a schedule when you run EVM on all those areas.

EVM absolutely does reference the schedule, remember that one of the benefits is that it integrates cost and schedule, it just uses costs (or whatever other resources might be loaded) to calculate the schedule performance.  That's what SV and SPI are (remember Schedule Variance and Schedule Performance Index).  The problem with EVM is the issue of fixed price.  If you are not calculating Actual Cost using true or real contractor job costs, you are not using the EVM the way it was originally designed.  But with fixed price jobs, Actual Cost is really just the percentage of the contract value that has been installed.  But the Earned Value is a time-related metric, it calculates SV and SPI.  One way that might help to think about it is the way that the old Primavera P3 worked, you had to either link or delink cost and time and if you linked them, you did not need to update the percent complete values and your SPI and CPI were identical.  But the percent complete updated the cost metrics and the remaining duration updated the schedule metrics.  I always said to our folks that we get the percent complete for activities from the subcontractor invoices (and in-house estimates of completed work value for self-performed work) because they were focused on how much work had been installed, and we get the remaining durations from the field supervision's estimate of how much work remains, or how much longer it will take to finish individual activities.  In order to get both cost and time metrics from P6, we have to update the two values separately, we can't get it from simply updating actuals and remaining durations.  I hope that helps.

Check the paper I uploaded about Duration Day EVM because it might help understand how the metrics work for fixed price contracts.  Also, see the attached papers by Hannah Schumacher and Charlie Jackson related to cost loading P6.  All of their papers on use of P6 are worth keeping in your library.  I'm also uploading a paper by Charles Lappenbusch which discusses problems with EVM implementation, and another good paper on how best to use EVM by Joseph Lukas.

I just worked on a recommended practice for planning packages and work packages and it was hard reconciling the differences between those who wanted strict compliance with EIA-748 (the purists as my co-author, Andrew Dick, calls himself) and those who felt like EVM is useful at a simpler level.

I hope this is useful, I suspect you'll get to hear a lot of alternate positions from others, and that's good even if it hurts my feelings!  LOL.

Chris, you are saying that the EVM curves will show a jump where the plan changes.  From the various posts it seems it is allowed to amend the plan once it is approved and only rebaseline when there is some major change.  Basically for any approved change to the cost and schedule plan, it is okay to update the plan to reflect the new time phase plan for remaining work.  Maybe the term re-plan and baseline are different.  The second question is how is EVM used to monitor float erosion?  Not sure that is clear. EVM doesn't reference the schedule and only deals with cost.

Michael:

You raise several issues, so let me take them individually.

1. Re-Baselining - whether you rebaseline or not, the EVM curves will show a jump where the plan changes.  I believe it is necessary to rebaseline when the plan becomes significantly different or the schedule will no longer be reliable.  And when they rebaseline, everything that is still in the as-planned stage will get reworked, that's normal.  What I don't believe is normal is for a contractor to try to rework the full schedule starting from NTP - actualized data cannot be changed.
   1. A rebaseline, by its nature, will formally change the schedule, that's to be expected.
2. SPI returns to 1.0 and SV returns to 0 at project completion.
   1. This is part of the downside to running EVM on fixed price jobs where the planned value is the sum or the price to the owner, and not the costs of the contractor.  When EVM (C/SCSC - Cost and Schedule Control Systems Criteria) was originally envisioned, it was only used for cost reimbursable projects and true job costs were collected as Actual Costs.  But over the years, people realized that EVM is useful for invoicing so owner costs were loaded, and the BAC can never exceed the PV (change management excepted).  This means that in the last 30% or so of the project, the time metrics will improve regardless of the status of the project, even with very late predictions of completion.   
   2. You should check out Walter Lipke, who invented "Earned Schedule" which computes how late the project is by comparing where the planned value curve crosses the data date compared to the earned value curve.  This is similar to how we looked at it in the old graphical EV days, and works quite well, even when the project is trending late.
3. Taking on Earned Value
   1. Be aware that EVM run at the project level is just not all that useful because an early delivery of a large piece of equipment can make the metrics look good even when the project is in trouble.  You still need critical and near-critical path analysis because those are the activities that will cause immediate delay or experience a mid-period critical path shift and cause delay.  EVM is good for trending analysis, similar to duration overruns, missed starts, float erosion, and other methods that look at performance but not necessarily critical path delay.
   2. But when running EVM on filtered data sets like trades, submittals, floors, processes, critical paths, it can very useful and provide good insight not easily gained in other ways.  You can run EVM on anything that has activity codes to isolate the type of work
   3. With fixed price contracting, the time metrics are more useful than the cost metrics (even though time is presented through cost).  The better you do with loading costs, the more reliable you will find EVM metrics.  I like to see large equipment packages, like service equipment, boilers, light fixture packages, loaded into one-day delivery activities. That way when the equipment hits the job, it can be actualized and added to the invoice automatically.
   4. There are several RPs that you will find useful and should take a look at, RP82R-13, 79R-13, 80R-13, to name a few.  Also the Library is chock full of good papers on EVMS.
4. We use a technique that I call "duration-day" earned value (same name used by Murray Woolf, the only other person I know who has run something similar).  One of my colleagues, who has done a lot of EVMS aligned with EIA-748 global standards, really didn't like my approach, so I challenged him to co-author a paper and we did so this past summer.  Our conclusion?  That DD EV is very useful and can be implemented successfully.  I like it because our schedulers can do it with global changes and we don't need contractors to put in any additional work. 
   1. I've attached our paper for your use.

I don't know if this answers your questions, and you might want to post it in the EVM Community to ensure everyone who is engaged with earned value monitoring sees it.  

i wish i were getting such quality advice in the past. So much information in compressed form. Books worth of it in a page.
It would have saved me months and years of being sidetracked by bluffers.

Michael:

Something that I thought I said but I don't think was clear, there is a large difference between running EVMS on fixed price projects compared to cost reimbursable projects, and I don't think that industry literature generally addresses fixed price contracts.  I suggested that you post in the EVMS community but remember that those folks are mostly the federal EVMS people, so they are fully attuned to following the EIA-748 standards for cost reimbursable projects where they are loading contractor's actual costs, which means that they rely on cost metrics as much as schedule metrics, whereas in the fixed price world, we tend to rely more on schedule metrics and recognize that there is value in the cost metrics but since those are simply percentages of the contract amount, they are most useful when the project is on time and budget.  Also, AACE and commercial/non-governmental best practices note that contingency is assigned to potential risks and is expected to be spent, and is visible to the project, whereas management reserves are hidden from the project and only expected to be spent for scope changes.  In the federal government world, that is reversed.  Just so you'll be aware.

I said that you can see a jump in the EVM curves if you do not re-baseline, so I think it is obvious and can be easily explained if you don't re-baseline.  In the fixed price world, from my perspective, I use the baseline and re-baseline cost loaded schedules only as a vehicle to hold the contract - meaning that it is my reference to contractually mandated cost and time.  So, if I have a change order that increases costs and/or provides an EoT, I would add an activity (or multiple if they affect different contractual milestones) that would carry the additional cost and time, and just link that at the end of the project.  That way when I reference back to the contractual re-baseline it gives me the right cost and time.  My most reliable comparison benchmark is the latest updated schedule as long as it is being reviewed and approved because that is the contemporaneous understanding of the critical path at each update.

But I believe one of the other large differences in running EIA-748 compliant EVMS for federal governmental projects is that the baseline schedule is of vital importance acting as the benchmark against which project performance is measured, so rebaselining is a harder task than in a fixed price non-federal project where you could just adjust logic in an update and compare it to the previous update.  In the fixed price world, I do not like changing logic in the baseline schedule to create a rebaselined schedule because the conditions are very different at the rebaseline point than at the original baseline.  I think it is dangerous to start looking at how the project would be planned differently in the baseline (at the time of creation of the baseline) because I think it leads to someone trying to determine "what the schedule would have looked like if it had been created to include these two years later changes".  I think that is what is dangerous because the field conditions are not the same and trying to recreate a new baseline that somehow is rooted in the original plan but now contains new conditions can be misleading.  I think it's better to keep the schedule revised to match the current plan and just compare the contract time and cost, while benchmarking against the previously approved "contract" schedule.  

But I think in the cost reimbursable world, (in my experience only) we tend only to rebaseline when there is a significant change to phasing or sequencing, or a significant change order, not for every change.  My perception is that most practitioners in that world keep the cost rebaselined but are not so fixed on the time.  Others will have more insight.

So, I think for practical purposes replan and rebaseline are reasonably the same for larger needs.  You could say that the contractor replans every time he updates the schedule and performs maintenance to keep the schedule serving as a good model of the work.  But I don't think most people rebaseline for every small change, but as I said, for me, the rebaseline is a current benchmark of the contract so I do rebaseline for every change order that has time or cost.

With respect to float erosion, remember that we have to monitor 1, the critical path since those are the activities that will immediately cause delay and prolong the completion predictions, 2, the near-critical path since those are the activities that might suffer a mid-period critical path shift, sneak up on the team (as they are focused more on critical paths) and cause a delay that also prolongs completion predictions.  But all the work must be installed at a rate commensurate with completing on time, and the way to monitor that best is with the various trending analysis methods.  Those methods include float erosion, missed starts and finishes, duration overruns, resource analysis, and earned value.  So, while EVMS doesn't measure float erosion (in fact, if you don't take into account the total float values, your EVM metrics will not be so meaningful.  The way that I merge the two needs is by running EVM on subsets of the schedule, such as the critical path, the near-critical path, the trades (to ensure that a trade not on the critical path is not allowing momentum to falter because he has lots of time before going critical), the various stages and phases, areas, and anything else with an activity code that allows filtering.  You might be surprised at how much more insight you get from a schedule when you run EVM on all those areas.

EVM absolutely does reference the schedule, remember that one of the benefits is that it integrates cost and schedule, it just uses costs (or whatever other resources might be loaded) to calculate the schedule performance.  That's what SV and SPI are (remember Schedule Variance and Schedule Performance Index).  The problem with EVM is the issue of fixed price.  If you are not calculating Actual Cost using true or real contractor job costs, you are not using the EVM the way it was originally designed.  But with fixed price jobs, Actual Cost is really just the percentage of the contract value that has been installed.  But the Earned Value is a time-related metric, it calculates SV and SPI.  One way that might help to think about it is the way that the old Primavera P3 worked, you had to either link or delink cost and time and if you linked them, you did not need to update the percent complete values and your SPI and CPI were identical.  But the percent complete updated the cost metrics and the remaining duration updated the schedule metrics.  I always said to our folks that we get the percent complete for activities from the subcontractor invoices (and in-house estimates of completed work value for self-performed work) because they were focused on how much work had been installed, and we get the remaining durations from the field supervision's estimate of how much work remains, or how much longer it will take to finish individual activities.  In order to get both cost and time metrics from P6, we have to update the two values separately, we can't get it from simply updating actuals and remaining durations.  I hope that helps.

Check the paper I uploaded about Duration Day EVM because it might help understand how the metrics work for fixed price contracts.  Also, see the attached papers by Hannah Schumacher and Charlie Jackson related to cost loading P6.  All of their papers on use of P6 are worth keeping in your library.  I'm also uploading a paper by Charles Lappenbusch which discusses problems with EVM implementation, and another good paper on how best to use EVM by Joseph Lukas.

I just worked on a recommended practice for planning packages and work packages and it was hard reconciling the differences between those who wanted strict compliance with EIA-748 (the purists as my co-author, Andrew Dick, calls himself) and those who felt like EVM is useful at a simpler level.

I hope this is useful, I suspect you'll get to hear a lot of alternate positions from others, and that's good even if it hurts my feelings!  LOL.

Chris, you are saying that the EVM curves will show a jump where the plan changes.  From the various posts it seems it is allowed to amend the plan once it is approved and only rebaseline when there is some major change.  Basically for any approved change to the cost and schedule plan, it is okay to update the plan to reflect the new time phase plan for remaining work.  Maybe the term re-plan and baseline are different.  The second question is how is EVM used to monitor float erosion?  Not sure that is clear. EVM doesn't reference the schedule and only deals with cost.

Michael:

You raise several issues, so let me take them individually.

1. Re-Baselining - whether you rebaseline or not, the EVM curves will show a jump where the plan changes.  I believe it is necessary to rebaseline when the plan becomes significantly different or the schedule will no longer be reliable.  And when they rebaseline, everything that is still in the as-planned stage will get reworked, that's normal.  What I don't believe is normal is for a contractor to try to rework the full schedule starting from NTP - actualized data cannot be changed.
   1. A rebaseline, by its nature, will formally change the schedule, that's to be expected.
2. SPI returns to 1.0 and SV returns to 0 at project completion.
   1. This is part of the downside to running EVM on fixed price jobs where the planned value is the sum or the price to the owner, and not the costs of the contractor.  When EVM (C/SCSC - Cost and Schedule Control Systems Criteria) was originally envisioned, it was only used for cost reimbursable projects and true job costs were collected as Actual Costs.  But over the years, people realized that EVM is useful for invoicing so owner costs were loaded, and the BAC can never exceed the PV (change management excepted).  This means that in the last 30% or so of the project, the time metrics will improve regardless of the status of the project, even with very late predictions of completion.   
   2. You should check out Walter Lipke, who invented "Earned Schedule" which computes how late the project is by comparing where the planned value curve crosses the data date compared to the earned value curve.  This is similar to how we looked at it in the old graphical EV days, and works quite well, even when the project is trending late.
3. Taking on Earned Value
   1. Be aware that EVM run at the project level is just not all that useful because an early delivery of a large piece of equipment can make the metrics look good even when the project is in trouble.  You still need critical and near-critical path analysis because those are the activities that will cause immediate delay or experience a mid-period critical path shift and cause delay.  EVM is good for trending analysis, similar to duration overruns, missed starts, float erosion, and other methods that look at performance but not necessarily critical path delay.
   2. But when running EVM on filtered data sets like trades, submittals, floors, processes, critical paths, it can very useful and provide good insight not easily gained in other ways.  You can run EVM on anything that has activity codes to isolate the type of work
   3. With fixed price contracting, the time metrics are more useful than the cost metrics (even though time is presented through cost).  The better you do with loading costs, the more reliable you will find EVM metrics.  I like to see large equipment packages, like service equipment, boilers, light fixture packages, loaded into one-day delivery activities. That way when the equipment hits the job, it can be actualized and added to the invoice automatically.
   4. There are several RPs that you will find useful and should take a look at, RP82R-13, 79R-13, 80R-13, to name a few.  Also the Library is chock full of good papers on EVMS.
4. We use a technique that I call "duration-day" earned value (same name used by Murray Woolf, the only other person I know who has run something similar).  One of my colleagues, who has done a lot of EVMS aligned with EIA-748 global standards, really didn't like my approach, so I challenged him to co-author a paper and we did so this past summer.  Our conclusion?  That DD EV is very useful and can be implemented successfully.  I like it because our schedulers can do it with global changes and we don't need contractors to put in any additional work. 
   1. I've attached our paper for your use.

I don't know if this answers your questions, and you might want to post it in the EVM Community to ensure everyone who is engaged with earned value monitoring sees it.  

Hi Mr Persaud.

Lots of writing from me, just to save you the back and forth later on. In order to leave the cluttered detail for later, go for bold and underlined first.

Quickly: regarding whether it is ok or allowed, not that it is allowed, but it should be required. Or, to put it in milder terms, it is a good and recommended practice. As for timing, it had to be the part of the change management process, and a pre-requisite for the approval of the change. It would be quite a surprise not to see it in the contract or technical specifications adjunct to the contract

You almost apologized for, quote, 'a basic earned value question", but it seems that somebody else may need to apologize. You need to do your work, and you are in the right.  You want to avoid being told later that "your numbers are useless" by the same.

More details below, after the information sources (italics).

As for sources, there are many, some of which listed below, where those few which will make a good start are in bold font. Others may be used when you need more detail.

That should have actually been the part and parcel of entire change process in the first place. A bit strange that the change was approved without such information being provided.  As stated above, some if not all of the steps should already be in contract documents.

If not, I would venture a guess that there must be something somewhere, an SOP or a Project Management Plan in some drawer/sharepoint folder, which describes (and probably mandates) such steps in change management process.

Sources

Without that, if you need something to back up and support your initiative, you can just go through the usual AACEI sources, such as RPs related to EVM in one or another shape. Off the cuff, here are the numbers: 82R-13, 81R-13, 80R-13, 79R-13, 75R-13, 86R-14.

These mostly also refer to EVMS per EIA-748.  I also remember that there were some presentations, papers and materials provided by individual AACEi members, such as Michael Nosbisch and Dan Melamed.

Not to be overwhelmed by the entire EIA-748 (where the guide can be downloaded from various DOD, DOE et al websites), it would be good to start with the 82R-13.  EIA-748 guide may be far too regimented for an introduction. (Regarding schedule specifics, the guide is complemented with PASEG, also to be found and downloaded from various NDIA-related web sites, like www.ndia.org)

If you prefer a visual example-based introduction, though still within EIA-748 core, there are short video snippets on DOE web site at 

https://www.energy.gov/projectmanagement/evms-training-snippets

If you need to skip introductions, you may go directly to the ones related to OTB/OTS (over budget, over schedule) ones, such as 1-5A, 1-5B and 1-12. Again, this is still very regimented and client specific, but you can get an idea about the most meticulous steps, and then apply it in your project environment (even if you do find somewhere there an SOP).  It is not bad to have a gold or platinum card, that is, EVMS (as per 748) cheat sheets.

Those can also be downloaded:

Gold Card https://acqnotes.com/wp-content/uploads/2014/09/DAU-EVM-Gold-Card-2020.pdf

Platinum Card is meant for Agile. In case you need it, just search for "Lockheed Platinum EVMS Card" or similar.

Back to the nature of your particular situation,

where more information may be needed to provide you with the right answer. In absence of that, I will start with assumptions. You pick what could be useful to you, if anything.

It appears that you may have been excluded from the change process (and that you may not have the support of some SOP). I am not quite sure what is your position, cost controller, scheduler or accountant.  I say that because I came across environments where time distribution of costs was created by cost controllers, accounting personnel, and even people holding titles of project manager and project controls manager, but without any schedules ever being created or cost/resource loaded. They just knew the start and finish dates. Not a  mom and pop outfit, but a large multi-national company, hundred or so years in the business, one of those which are part of the lore (such line always reminds me of Barings Bank and their spreadsheets).

Were your "time phased values" produced based on a cost/resource loaded schedule? If not, it would be of help to create one for this purpose.

On the other hand, if you do have a cost/resource loaded schedule, you certainly have to apply that change in a re-baselined schedule (which should have been part of the change approval process, as already stated). All properly documented, changes tracked and listed. 

Some other points which you may need to deal with:

-Who handles management reserve and contingency. How will you address those?

-You mentioned that the past portion of the plan was also changed. How did you handle that till now, in particular with SPI/CPI when it comes to possible past deletions or additions of cost/resource loaded activities?

Hope that at least something of the above helps, be it with the details of the work or regarding your initiative with higher-ups.

------------------------------
Boris Konrad
CCP, PSP, PMP, PMI-SP, PE
United States
------------------------------

When a company implements Earned Value they write their own rules normally based on a standard or recomended practice. If you have not contract or corporate requirements then there are no rules for you, but obviously some guidelines.

It is my opinion that the baseline should be reset when the last baseline is irrelervant. i.e. It may not be seen on a Gantt chart because of delays or the plan has been significantly changed, whch there appears to have happened in your project.

When a baseline is reset then variances before the rebaslining are calculated on the old basline and variance after rebaselining on the new baseline.

I feel a major disadvantage of not rebaslining is that the project manager has a target that may not be met and the motivational effect on the project team of a meaning ful target is lost.

It is my opiion that rebaselining is one of the grey areas of Earned Value and in some EV documents is not well documented or not documented at all. 

The  APM Guidelines state changes may only be made to the baseline for:

• Authorised changes to the scope, cost or schedule of the project,
• Changes to standard rates (e.g. labour, material or overheads),
• Work and budget transfer between control accounts.

I often ask ChatGTP question about planning and scheduling and usually I receive unhelpful answers. For example I asked what were the advantages and disadvantages of P6, Microsoft porject and Powerproject and receive absolute garbage. But I was supride by the answer I received and I feel it is not too bad:

My Question to ChatGTP

------------------------------
Paul Harris
President and Managing Director
Eastwood Harris Pty Ltd
Templestowe VIC, Australia
0411187701
------------------------------