Hi Bernard,

Just wanted to pass on the news that Micro Planner X-Pert enables PERT diagrams to be draw with "Activity on the Arrow" (ADM) as well as PDM.

Cheers
John

Bernard:

I think we are making progress.

You are saying "In the diagram presented, ES(2,5) [the dummy] = ES(2,3) [framing] because both are dependent upon E(2)." E(2) is the early time for node 2. Note that it says: dependent upon E(2).

Then you are saying "... Any delay on the ES of the framing activity is a delay on the dummy ...". This is diferent from the above paragraph. Any delay of the dummy is caused by any delay of node 2, not by any delay on the start of activity 2-3. These are two diferent things.

Or, if we put it in another way, where in your referenced book does it say that the early start of any activity is calculated from the early start of any other activity bursting from the same node?. Note that I am not saying the early time of the node, but the early start of the activity.

And also, why not the other way around? Why the start of activity Frame Wall is the one that is dependent on the start of the dummy activity? or for that matter any other activity that bursts from the same node? If we had 10 activities bursting from the same node, which one would determine the start of others? Or any delay on any of the activities will delay the start of all the others? What a beauty!

Tomas Rivera
Altek System
Detailed scheduling and control of
high performance construction projects.

Bernard:

Let me give it one more try. I went to my library and looked for an old book Precedence and Arrow Networking Techniques for Construction, 1978 by Robert B. Harris. All the quotes used will be references from this book. I will demonstrate that the start of activity Frame Wall does not determine the start of activity Rough Piping (or the start of dummy activity 2-5 for that matter).

I am going to start by saying that the fact that if two activities have equal early start date, it does not mean that one determines the other. It does not mean there is a start to start relationship.

Let us state some basic concepts for the ADM (Arrow Diagraming Method, or Activity on the Arrow method) method. "... the activities fell on the links (arrows) and the interrelations between activities were assigned to the nodes." This is saying there are two distinct elements. Nodes and arrows. Nodes are not part of the arrows. Therefore nodes are not part of the activities. Therefore, node 2 in the example mentioned numerous times, is not part of the activity Frame Wall. It only determines the earliest time it can start. If the activity is identified by the node numbers, it is only derived from a naming convention.

"In the forward pass computations the maximum of the EFD (early finish dates) values for all activities merging at a node was taken as the single ESD (early start date) value for all the activities that burst from the same node. This fact allows the assignment of this maximum number to the node itself as the early time of occurrence for the event that marks the beginning of the following activities." From this statement you can see that the early start date of any activity is determined by the latest early finish of the activities merging at the beginning node of the activity. The start of any activity does not depend on the start of any of the activities bursting from its beginning node. Using the example mentioned, the start of dummy activity 2-5 does not depend on the start of activity Frame Wall 2-5, its start depends on the latest finish of the activities merging into node 2. To be more specific, the start of Rough Piping does not depend on the start of Frame Wall.

I can go on citing statements and calculations from the referenced book that show the same conclusion. If you make the computations using the example arrow network with durations for everyone of the activities equal, for example one or two days for each one, you can see that the late start date for activity Frame Wall would be at a later date than the late start date for Rough Piping. This should show you that the start of Frame Wall does not determine the start of Rough Piping.


Tomas Rivera
Altek System
Detailed scheduling and control of
high performance construction projects.

Bernard:

In a way you got me and I am sorry. Let me explain and expand my point.

If the job advances: your deck is poured, and we have a strike of carpenters that prevents activity Frame Wall to start. At this point you update your schedule, you do not know how long the strike is going to last, but you know you will have a delay from this point forward, and you need to take this into account somehow. You decide to reflect a tentative delay of one month to the availability of the resource carpenter. To be true with reality you need to do just that: show a delay of one month to the availability of this resource. The proper way to do this is to adjust the calendar of this resource (I acknowledge that this, in a way, is a constraint). When you do your calculations, the activity Frame Wall will try to be scheduled when there are resources available (you need to do resource leveling in order to represent the reality of your jobsite). Since there will not be resources available for one month, this activity will be delayed one month. Now to the main point of discussion: the activity Rough Piping will be scheduled for inmediate execution since there is no problem with its own resources and its only predecessor (Dummy 2-5 that comes from the finish of activity Pour Deck) is already completed.

Now, we might have a problem in modeling this exact reality if we do not have a software that has the proper features like resource calendars or resource leveling. If we do not have this feature in our software we need to resort to something equivalent that gives us the same result. This could be done by adding a predecessor to activity Frame Wall named Carpenter Strike with one month of duration, and you will obtain the same results as in the previous paragraph.

I hope I made a clear explanation and did not make a statement that might not be true in every condition. In other words, I hope that the main message is understood.

Tomas Rivera
Altek System
Detailed scheduling and control of
high performance construction projects

Bernard:

I hate to say it but, I agree with Ronald almost in everything he has said right since his first post.
The main point Ronald is stating, is that Frame Wall does not determine any date for Rough Piping. Frame Wall´s predecessor does, but that is a whole diferent situation although calculation gives you same dates in the case of no delay. If you delay activity Frame Wall in the ADM model (without changing logic o adding constraints, of course), activity Rough Piping will not be delayed; hence, you will have diferent dates.

It is possible, I do not rule this out, that there is a communication problem here, as you implied. Ronald and myself might be looking at the same side of the problem, and you might be looking at another side of the same problem. And we do not get to understand each other. But I find this possibility very unlikely in this case.

I would be eager to have someone to demonstrate my mistake in this seeming simple and basic scenario, but of great consequences.

Tomas Rivera
Altek System
Detailed scheduling and control of high performance projects

Lets just say that I disagree with your previous statement. In the ADM example, Framing is a predecessor to nothing, no matter how much you want it to be.

I must be misunderstanding what you are saying. Are you saying (using my ADM example below,) that framing is the predecessor to rough plumbing? It is not. Pour Deck is the predecessor to both framing and rough plumbing. Delaying framing has no effect on the start of rough plumbing.

In a PDM situation using a SS relationship from Framing to Rough Plumbing, you statement would be correct (which is my point.)

Construction delay claims is such a precise issue that it is very difficult to say exactly what you mean in short, concise statements that apply to all situations.

In the event of a Carpenter’s strike, you would be allowed to add the constraint to the framing activity. What you would generally not be allowed to do is to say, “Everyone knows that the plumbing would also be affected by the strike so I am going to add an additional constraint there as well.”

The logic in the schedule did not support this second modification. Perhaps you never did intend to wait on the carpenters before starting on the plumbing and now you want to take advantage of this situation to hide your earlier mistakes and delays.

In this case, ADM says no delay to plumbing, and thus no project delay. PDM would have shown a logical reason for the Carpenter’s strike delaying the completion of the project, thus excusable and perhaps compensable extension to the project.

Tomas,

Thanks for joining in!

I still do not understand how a PDM system is scheduling the network any differently than an ADM system. AFAICT, both systems will calculate the task availability exactly the same. Both systems will calculate the successor task’s ES to be equal to the predecessor’s ES in the absense of an overriding relationship. If the predecessor slips, the successor will slip as well.

...

OK. I just had an epiphany. Ron, I think you are talking specifically about an arrow diagramming chart versus a precedence diagramming chart for displaying a logic network versus the schedule calculations / Gannt charts that would result from calculating either network? In that sense, I would agree that the listed ADM structure is not an exact representation of the PDM SS relationship. However, the practical effect of using either system would be as I described (equivalent calculations).

PDM charts which only use FS & SS relationships can be described equivalently with an ADM chart if all relationships are described using dummies, but this would not be practical for humans.

Bernard Ertl
InterPlan Systems Inc. - Project Management Software, Project Planning Software

Let us look at this logically and theoretically.

1) ADM only uses Finish-to-Start (FS) relationships. Period. There is no other relationship allowed. When all predecessors to a node are complete, then all successor activities to that node can start. You cannot gauge logically when an activity starts, only when the preceding node has all predecessors complete.

PDM, on the other hand allows you to logically gauge when the activity starts (via a Start-to-Start (SS) relationship.) It has more functionality than the ADM ‘equivalent.’

2) ADM also does not have a facility for logically overriding the CPM calculation based upon when an activity finishes. You can not tell an activity to finish only after some other activity has started or finished.

PDM has such a facility with the Start-to-Finish and the Finish-to-Finish (FF) relationships. You can logically delay the start of an activity by defining when it may finish.

Will you concede that if two things exist and one of them has more functionality, then they are not equivalent? The difference may not appear important to you, but believe me, it is important in claim situations.

What a great idea; simply post a constraint where the logic fails to accurately reflect the real-world consequences. We could do that on all activities and eliminate all of those pesky relationships that only get in the way anyhow.

Seriously, there is nothing wrong with ADM, especially if you do not care about the legal ramifications that I was speaking of. Unfortunately, the entire construction industry is unable to simply add constraints or even change the logic in these situations. It is important for Construction Schedulers to understand this before it happens to them instead of after and wish that they had.

Your experiences with your study and findings would make an excellent article for a serious journal. Have you considered writing it up and submitting it to the Association for Advancement of Cost Engineering (AACEi)? I just looked it up and the Editor of their journal (Cost Engineering) is Marvin Gelhausen, e-mail: mgelhausen@aacei.org (phone: 800.858.COST ext.: 111).

Ron,

I’m still not sure I’m following you. With either ADM or PDM, and the scenario you laid out, completing the two predecessor tasks (Pour & Deliver) would allow both the framing and piping tasks to be scheduled to start immediately on the timeline. In the absense of any calendar constraint on the framing task (to indicate a delay such as a strike), both tasks will continue to be scheduled as available starting on the timeline (through successive updates) until progress/completion/rescheduling occur. If you introduce a calendar constraint on the framing task, the piping task will be affected as well.

Bernard Ertl
InterPlan Systems Inc. - Project Management Software, Project Planning Software

Ron,

[quote]In certain cases, you will calculate out the same early start days and in other cases, it will be different.[/quote]

Could you provide an example where the calculation will be different?

Any delay on the start of the predecessor will delay the start of the successor in the arrangement I described.

Bernard Ertl
InterPlan Systems Inc. - Project Management Software, Project Planning Software

Ron,

Thanks for being willing to discuss the topic.

I’m not sure I’m following your argument. In a PDM construct, the early start (ES) of the successor task will equal the early start of the predecessor task (in the absence of another overriding relationship). The calculation is literally:

ES(k) = ES(i) + SS(i,k) for preceding task i with a SS relationship to k

The same exact calculation is achieved with the start node dummy between two tasks in ADM.

As far as I can tell, the only difference is in the terminology used to describe the relationships, not the actual implementation.

Bernard Ertl
InterPlan Systems Inc. - Project Management Software, Project Planning Software

Ron, IMO, ADM does offer a direct equivalent to SS relationships in PDM networks through the use of a dummy activity/tie. I’m not sure if you were being a semantic purist in your post (ie. no direct equivalent without dummy ties), but an ADM network can offer a direct equivalent to a PDM SS tie (that acts the same as a PDM SS relationship).

I posted a simple comparison of ADM & PDM scheduling constructs in this Critical Path Logic thread.

Bernard Ertl
InterPlan Systems Inc. - Project Management Software, Project Planning Software