CCEVM Evaporating Cloud Diagram
A Goldratt technique came in handy to map out where I see the conflict between CCPM and EVM that I referred to in my last post. Please excuse it’s sloppiness, I will try to make a cleaner computer-generated version later on, but I think I may be refining this later on anyway.
I’ve never seen an evaporating cloud with more than 1 requirement for each of the prerequisites, but I found it necessary to have a requirement which stems from both CCPM and EVM. The conflict indicated between employing Critical Chain and EVM stems from different behaviors being driven by the two. Critical Chain supports behaviors that focus on efficiency with tasks on the Critical Chain , thus improving the outcome of the project. EVM supports behaviors that make it appear overall cost efficiencies are good, even if those efficiencies are being achieved on tasks that aren’t critical to the completion time of the project. Project managers might decide to work on some easier non-critical tasks if their EVM is going to fall short and get a short-term EVM win, but if that happens it throws EVM’s predictive power regarding schedule out the window.
The resulting direction from this is to modify the budgeting and cost control tools in the Critical Chain body of knowledge. It needs to use buffer management methods for cost, to control the project that is implemented in such a way to make it compatible with existing EVM metrics. There would be a single project cost buffer which is already part of the CCBOK. Cost buffer management would be used for controlling project costs, in addition to an accurate EVM translation based on cost buffer utilization compared to planned utilization.
Note: from a cost perspective, all tasks are on the Cost Critical Chain (CCC) because cost over run in any task will make the project over budget unless other tasks have under runs. It makes no difference if they are on the schedule’s critical chain. That’s why there’s only 1 cost buffer, the project cost buffer.
Jan 16th, 2007 at 1:30 pm
Agree with your comments about EV and Critical Chain. The two can work together, but you need to be careful. Use Critical Chain to drive the project and be careful not to let EV distract your focus away from the critical chain/path.
Reply
Jan 16th, 2007 at 8:48 pm
I agree completely, and hope to come up with a solution that makes EVM metrics a direct result of CCPM methods, so they both reinforce the same behavior.
Right now EVM and CCPM are so disjointed, and they have the possibility of resulting in conflicting behaviors.
Thanks for the comment!
Reply
Oct 14th, 2008 at 8:42 pm
Josh — your description of the imperatives of EVM is simply wrong. You cannot create cost efficiencies by working on non-critical activities. If you work on an activity before you planned to work on it, you will show a positive schedule variance. If you complete that work for less than you planned, you will show a positive cost variance.
If your management is willing to accept positive variances as evidence of “goodness” without asking for an explanation, then either someone hasn’t taught them what they need to know about EVM, or you have ignorant managers and it really doesn’t matter what reports you show to them.
EVM is based on reporting against a baseline, and the assumption is that some thought was put into that baseline. So if you are working on something ahead of schedule, management should be asking “why?”
Don’t blame the tool …
Reply
Feb 18th, 2009 at 3:20 pm
There is no conflict is using EV reporting (which is what is typically “required” by policy) and using CC PM to plan and manage projects. There are published papers from organizations that did exactly that while trying out CC PM. In one published case at Boeing, they stopped paying attention to the EV reports once they realized the CC buffer management reports gave them better and earlier insight to problems on projects. If someone requires you produce some report, give them the report! So in your cloud there is no conflict between D and D’. The “conflict”, if any, comes when folks reading the EV report ask you what you are doing about the “problems” it shows. The actions you take using CC PM are perfectly reasonable actions even under the old EV paradigm.
Also, the goal in your cloud is not really just to make certain info available is it? Isn’t it to manage project better? Or bring project in as quickly and cheaply as possible? Your requirements (B, C, and F) aren’t necessary for “accessibility” but rather for “superior project management”.
And one last thing: how do you judge “optimal” cost or schedule performance? There are many case studies showing how CC PM can yield shorter projects with no cost increase. I don’t know if CC PM is “optimal”, but shorter with no cost trade-off is clearly better, yes?
Reply
Josh Nankivel Reply:
February 18th, 2009 at 7:17 pm
Thank you Richard. I would definitely enjoy discussing this more with you. I wrote this over 2 years ago and frankly, it’s difficult for me to understand fully what I was writing at that point.
Since then, I’ve written on a method to use EVM with Agile methodologies, but never again on CC with EVM. I’ve had some contact with Larry Leech on the topic, but I have yet to come into a good body of knowledge regarding CCPM and EVM. Most of the time, CCPM practitioners will say that EVM is basically a waste of time, and talk about ways of being able to check the box with EVM by backing into a report.
Reply
Feb 18th, 2009 at 10:55 pm
Bill,
Your description of working out of sequence was common practice in many domains. Josh is NOT wrong in describing these activties. Cost “gaming” is done in many places when subcontarcting of non-interdependent workpackages are used. I’ve experienced them persoannely in the DOE and used them to short term advantage myself. A Work Authroization System is mandatory on any credible EVM program. DOE mandates this in DOE 413.3 as does NASA.
The size of the variance must pierce the reporting barrier for questions to be asked. It would be unheard of to have no variance on a monthly CPR or NASA 533M, so “If your management is willing to accept positive variances as evidence of “goodness” without asking for an explanation,” needs a clear and concise domain and context explination to be of any use to the PP&C staff. By this I mean positive and negative variances are common occurances. It’s when they pierce the planned variance boundary – usually forecast from the previous perid – that review is needed.
The current EVMS System Descriptions found in DCMA audited sites prohibits the performance of out of seqeunce work. Performing such work creates unfavorable indicies in two ways:
1. ACWP is applied to zero budgets – no BCWP for the ACWP to go against
2. BCWP is applied to unopened work packages again creating unfavorable indicies
Nearly every DCMA EVMS System Description I have seen, worked with, or written prehibits such behavior.
As a final note, rolling up an individual task’s EV to the performance report is rarely done. Work Package EV is usually all that goes to the finance system. At the task level every firm we’ve worked with uses 0/100, 50/50, or apportioned milestones. The common cost system CostView defines and restricts this usage. In one major prime contractor no other performance measures are allowed. Most certaintly some incremental BCWP provided by the CAM.
At the WP level the individual tasks are aggregated and weighted to produce the BCWP. SPI is reported on the schedule side and CPI catches up usuall one reporting period later.
Reply
Feb 18th, 2009 at 10:59 pm
All combining CC and EV is common on IMP/IMS based programs. The CC aspects – buffer size – are derived from the Monte Carlo simulation of the needed “protection” for the deliverable date.
On a recent program the EV reports are stil used by Finance and Business Ops, but the “margin errosion” is how the CAMs managed on a weekly basis. A Monday morning meeting review the physical percent complete from the previous week (spacecraft construction and test), and the margin errosion below the planned level for that point in the program became the center of discussion.
As well Late Start/Late Finish reports and a thing called “bow wave” was used to guide that week’s work to “get to green” for the month end CPR.
Reply
Feb 18th, 2009 at 11:02 pm
All,
The final concept on EV intensive programs is to assess the SPI on the critical path and the probabilistic critcal path. While Josh is correct in that all tasks (Gawd let’s hope you’re really at the Work package level of EV) impacts the agregated numbers, the the CP numbers that require management attention. Over/Unders on non-CP elements are interesting to bean counters. The PM (Program Manager) wants to know if we’ll make the delivery date – launch dates are booked 18 to 24 months ahead. MR hold back can cover the standard deviation values of CPI.
Reply