Welcome to Planning Planet, checkout the Guild of Project Controls Knowledge CAR

Do you understand links in a PDM Schedule?

13 replies [Last post]
Patrick Weaver
User offline. Last seen 1 week 5 days ago. Offline
Joined: 18 Jan 2001
Posts: 221
Groups: None

Many schedulers build PDM networks that calculate the effect of a change incorrectly! Use the wrong combination of links and the effect of shortening a critical activity is a delay in the overall schedule. 


This post looks at five logical constructs that generate illogical outcomes from a changed duration:  https://mosaicprojects.wordpress.com/2016/06/12/critical-confusion-when-activities-on-the-critical-path-dont-compute/


Sagar Hassan
User offline. Last seen 9 weeks 1 day ago. Offline
Joined: 12 Dec 2011
Posts: 8
Groups: None

Dear All,

Please let me know how to print activity links in Primavera 6 as my management want to see link for all activities.


Sagar Hassan

Rafael Davila
User is online Online
Joined: 1 Mar 2004
Posts: 4131

To say the reality of PDM scheduling and the legally accepted definition of the Critical Path are in conflict IS WRONG because:

1) You can model contiguous as well as non-contiguous durations if using P3, a PDM scheduling software, technically there is no such conflict.

2) Some activities shall never be planned as of non-contiguous duration such as a de-watering activities, concrete pouring activities and concrete finishing activities, to name a few. These activities shall be modeled using contiguous duration rules; you cannot model contiguous if using non-contiguous as you will lose control on the activity duration because at any time it can become non-contiguous, therefore to prohibit contiguous duration models across the board on these activities is non-sense, too much on delay claim paranoia. 

3) Activities that can be of non-contiguous duration shall be modeled as such; you can model these activities as of non-contiguous duration if using contiguous model simply by splitting the activity; easy, under control and transparent. This approach is a better approach as it is more transparent, it exposes true intent on whether scheduler needs the activity to be contiguous or to be split and how. It also can disclose some necessary float metrics as to better understand the implications of changes in activity duration when critical as well as when not critical. The need for control and transparency is not hard to understand. 

4) Any software that cannot model such contiguous activity rules, even in the case of SS & FF links between two activities, shall be avoided. 

5) Contiguous Activity duration rules are not in conflict with critical path definition, in some cases it is a must, in other cases a better choice, in other cases a poor choice. These are rules any scheduler shall understand, you just use the applicable rule at the activity level. Global non-contiguous rules for activity duration as implemented by defunct P3 are non-sense unless used exclusively for some comparison purposes, in a similar way some "banned" constraint types were intended to be used. Unfortunately some schedulers insist on miss-using such global settings. 


6) In our courts, at least in my jurisdiction, schedules are considered incomplete if there are resource constraints not being considered. Ironically most CPM software such as MSP and P6 cannot yield reliable values of resource leveled float. The hands down approach is to use the model to disclose cause and effect. Each party presents its own case and here is where the conflicts begin. In some occasions a contiguous activity duration is appropriate in others it is not, usually it is a matter of contractual conditions and risk allocation. Placing too stringent conditions on how the responsible for the means and methods can be seen as a no-damage for delay clause, usually enforceable if obvious, usually not seen with good eyes. A software developer that advocates too much for such restrictive conditions would not be welcomed by construction contractors. 

Most Spider Project schedules, if not all, are aware of the issues with contiguous and non-contiguous models, it is clear within the appropriate float values, most of which can be graphically displayed. So obvious!

With regard to the statement that most lawyers and tribunals will never understand the issue.  I do not agree, the rules are easy to understand. I do not believe it takes much brain as to realize what will happen if an activity is broken into several segments or not. Perhaps the use of expert witnesses just add to the confusion. Usually here most cases are solved by using common sense the judge can rationalize in his determination. 

1. If the dependencies in your examples are set properly then they are not ladders.

2. There are many definitions of Critical Path. I don't know if zero total float definition is legal in your country.

3 and 4. Maybe but was not discussed here.

Patrick Weaver
User offline. Last seen 1 week 5 days ago. Offline
Joined: 18 Jan 2001
Posts: 221
Groups: None

There are two very distinct outcomes from this discussion

1. There are some specialist options in a limited number of tools that mitigate or avoid the problems highlighted in the paper (but there always have been - ADM 'Ladders solved the problem in the 1960s).

2. The reality of PDM scheduling and the legally accepted definition of the Critical Path are in conflict.

3. Most schedulers seem blissfully unaware of the problem. 

4. Most lawyers and tribunals will never understand the issue!


Rafael Davila
User is online Online
Joined: 1 Mar 2004
Posts: 4131

While Miklós warns about how added PDM functionality can complicate things to the unwary scheduler he further expanded the functionality of traditional PDM discrete links to a single type, he is embracing them all plus more. 


Any of the constructs can be modeled using different rules for activity duration [fixed/intermittent/elastic] as exposed in his paper. Miklós is not calling them wrong but advise schedulers to make sure they are aware and understand which rules are used and how are implemented by different software. 


Page 165 of this preview give us some of Miklós recommendations.

6- Use the activity-splitting-not-allowed approach since this is the original PDM assumption. The non-splitting approach provides you information on the early and late accomplishing of the activities, while the activity splitting approach usually does not.
7- If it is inevitable, then the activity splitting approach can be used, but remember that after this you make connections between the *start event’ and ’finish event’ of the activities and not between the activities themselves.
8- Be careful in case of disputes or when working with other schedulers. First you have to check whether your ideas about PDM are concordant.
These rules help to prevent the schedule from becoming confusing or incorrect.

My approach is to use activity-splitting-not-allowed and if inevitable to split the activity into individual contiguous activities keeping control at the individual activity level.

A single rule for calculation of activity durations as well as a single rule for resource leveling splits is not good enough, not all activities require same rules.

 photo NGE023_zps3kyx5mkz.jpg


Knowing activity flexes and super floats allows managers to optimize resource assignments on specific activities and to understand if activity performance can be interrupted and for what duration.

 photo FLEX0256_zpsmriuqbmo.jpg

Tom Boyle
User offline. Last seen 19 hours 50 min ago. Offline
Joined: 28 Nov 2006
Posts: 93
Groups: GPC Qatar


"Uninformed" is not a term that I would normally associate with either Vladimir or Rafael.

I think your paper - and the comments here - do well to highlight the pitfalls of non-FS links in PDM schedules.  Certainly the existing dominant tools in the US (P6 and MSP) are susceptable to them, and it would be good for some alternative approaches to gain traction.

I wrote a similar article to yours - mainly focused on application in MS Project and using the BPC Logic Filter add-in: http://wp.me/p6CCB4-3H.  I'm not sure that most of these structures can be reasonably avoided - especially when attempting to model large-scale linear construction.

Rafael Davila
User is online Online
Joined: 1 Mar 2004
Posts: 4131
  • Under the special case there are competing starts and finish constraints on an activity contiguous scheduling the activity will delay the activity as to meet all constraints. The intent is clear to those who understand the calculation rules for contiguous activity model. 
  • Under interruptible scheduling it will be scheduled intermittent but without disclosing the details of how each segment is assigned, without disclosing into how many segments the activity split, without disclosing the minimum period of a segment of work, and most importantly how much of the activity may not be performed until the predecessor is complete. Calculations rules that might be different for each particular activity. Under interruptible activity scheduling transparency is lost.
  • The normal scheduling of construction activities by contractors and subcontractors is for activities to be scheduled contiguous. Usually it is desirable to schedule construction activities as a continuous operation because intermittent work is less efficient, stopping and re-starting of an activity might mean re-mobilization of crews, the inefficiencies created might mean the duration of the segments might have to be longer than the duration of the continuous operation. Contractors and Subcontractors who bid the lowest and make the most in profits demand that the Continuous Mode be used to schedule most of their work.
  • There are cases when it is desirable to schedule the activity on intermittent segments because of what is known as reverse logic. Under reverse logic if you split the activity it might reduce the total job duration or because some work will satrt earlier it might provide some protection to prevent the activity becoming critical or delaying other activities. If it makes sense to split the activity into several segments it should be applied only when needed and under a controlled and transparent split where you would know the scheduled start/finish, duration and float of the individual segments.
  • In rare cases it might be better to fully stretch the construction activity and reduce workload of resources.
  • Usually it is the contractor's choice to use the means and methods of his choosing and decide how to schedule his own work as long as it meets contractual conditions. If by doing so he delays his job then he must absorb the consequences. If the client delays the schedule and wants to force the contractor into a less efficient plan then he must absorb the consequences. 
  • To mandate all activities to be scheduled under interruptible calculation option because it is possible one of the schedulers do not understand what it means is self-serving. All schedulers are supposed to understand the implications of each choice. 
  • Beware of unscrupulous team members that insist on one sided rules always to their favor.
Shanmugam k
User offline. Last seen 29 weeks 5 days ago. Offline
Joined: 26 Jun 2016
Posts: 2
Groups: None

Dear Engr,

Good Day,

My question is how to keep relationship while copy the activity in the same WBS.

for ex Activty A have a relationship more than 50 links to other activity with the different WBS, i need to copy a activity A mean whilr i should keep the existing relationship.

awaiting for your valuable reply




Rafael Davila
User is online Online
Joined: 1 Mar 2004
Posts: 4131

Same as Vladimir I don't see the problem in the fact that there are many ways activity duration changes may influence project duration.

  • All of the constructs outlined in the reference will produce correct outcomes for PDM contiguous activity durations, that someone does not understand how the contiguous model works is an issue not everyone have. 
  • To the untrained person the outcomes may seem counterintuitive but are nevertheless true based on PDM calculation rules for contiguous activity durations.
  • The constructs Miklos call them the traditional six types of critical activities. Their understanding shall not be a big deal as the constructs are not resource leveled and as the constructs do not include the maximal lag relationships Miklos have implemented in his models.
  • https://www.linkedin.com/pulse/reverse-critical-activities-miklos-hajdu?trk=mp-reader-card

Old Primavera P3 PDM Networks had some rudimentary Schedule/Level Calculation Option that was applied across the board to all activities Schedule Durations but Primavera P6 abandoned this option in favor of a contiguous activity model.

 photo Contiguous_zpsk5xpdbgv.jpg

  • As shown above it follows PDM Networks have no problems with the creation of interruptible activity models, but these have been abandoned as any activity interruption can be simulated using contiguous models by simply splitting the activity in a controlled, predictable and transparent way.
  • http://www.ronwinterconsulting.com/Making_CPM_Transparent.pdf
  • Interruptible activities can show incorrect total float and may even appear on the Longest Path when no such criticality actually exists. As a side point, it would also be nice to exclude interruptible activities on the Longest Path if they are not really part of the controlling work.
  • In the USA 99% of agreed schedules use the contiguous model.
  • Primavera P6 and Microsoft Project popular in the USA use the contiguous activities but they do not allow for interruptible activities. 
  • http://www.ronwinterconsulting.com/Reviewing_Resource_Leveled_Schedules.pdf
    See page 29.
    P6 does not have an interruptible activity setting. This means that the schedule cannot split activities and their resource requirements to allow for more precise resource allocations.
    • The resource leveling process cannot adjust activity durations, thereby adjusting the daily resource requirement.
    • The activity Total Float and Free Float values displayed are incorrect for resource leveled delayed activities and their successors. A better reference is found in the Remaining Total Float column but even this does not fully reflect ‘resource float.’  
  • To say that contiguous models and PDM schedules such as those created using P6 are no good to present court claims is new to me although it makes some sense because the issues in calculations of Total Float. 
  • It is a fact that most of today scheduling software are based on the contiguous model, it is a fact most are based on the PDM Network and the Activity on Arrow representation is for practical purposes an extinct dinosaur.

Patrick, with the current data certain activity sequence determines project duration and constitutes critical path.

DRAg shows what happens if the critical activity duration becomes smaller. Negative DRAG indicates that project duration increases, positive DRAG shows that it decreases.

In any case with the current data certain activity sequence is critical, with other data it may not. I don't see why some data changes determine criticality. Critical Path may change if some activity duration is changed, calendars change, resource allocations ,change, mandatory constraints change. It does not make current critical activities non-critical.

Criticality may be discussed not from the data changes point of view but considering activity delays. If activity delay increases project duration then an activity is critical.

Patrick Weaver
User offline. Last seen 1 week 5 days ago. Offline
Joined: 18 Jan 2001
Posts: 221
Groups: None

I'm rather surprised by these ill-informed comments - the fundamental definition of a critical path is 'the sequence of activities that determine the earliest possible completion date for the project or phase.'  If the activity takes more time to do but the schedule takes less time to complete, surely there is a problem n the PDM methodology. 

the paper is not academic - on a number of occasions as an expert in court, I have destroyed claims by the other side simply by using the schedule agreed under the contract to show the alleged delay to an activity in the schedule had no effect on the project completion and therefore there was no compensatable delay to the project. 

I know sophisticated tools such as Spider can avoid the problems and agree with Rafael that the calculations are entirely predictable but neither of these observations changes the facts for 99% of the project schedules in existence today.


Rafael Davila
User is online Online
Joined: 1 Mar 2004
Posts: 4131

All of the constructs outlined in the reference will produce the right information, that someone does not understand how the contiguous model works is an issue not everyone have. The outcomes are correct and very logical indeed and are what shall be expected of a contiguous model. The contiguous model is superior because it gives the user more control.

For lag drag there is no such thing as a single solution that fits all. 

Lag DRAG photo LagDrag_zps0hlkfxek.jpg

Automatic rules to solve the "lag drag effect" that occurs when an increase in activity duration reduce start of a successor issues are wrong.

  1. In particular when the rule is automatically applied to all activities.
  2. It makes no sense to split, to apply a reduced workload or a reduced production rate to all non-critical activities showing lag drag as in most such cases it is better to keep the activity execution contiguous. More frequently than not a reduction in labor production rates does not makes sense, in any case a workload reduction makes more sense, humans and machines are different. 
  3. Even if critical it might not be wise to split, reduce the workload or the production rate to the activity as other solutions might be better.
  4. Automatic splitting, automatic workload reduction and automatic production rate reduction are three different options that might hide the occurrence of Lag Drag and the possibilities of better solutions.
  5. Lag Drag can be identified and remaining lag drag quantified very easy by looking at the appropriate measure of flexibility, in Spider Project this would be DRAG and Start FLEX.
  6. It might be that not only critical activities lag drag justifies a managed split, a reduced workload, a reduced production rate or a combination but also it might apply to near critical activities with high probabilities of becoming critical. Figuring out which near critical activities have high probabilities of becoming critical is not addressed by simple deterministic models.

Lag Drag is an issue to be addressed when present and only when justified the activity split, a reduced workload or a combination shall be executed in a controllable way, easy solution is wrong, not a good approach.

Well, I don't see the problem in the fact that there are many ways activity duration changes may influence project duration.

DRAGs show this.