(1) Explain your problem, don't simply post "This isn't working". What were you doing when you faced the problem? What have you tried to resolve - did you look for a solution using "Search" ? Has it happened just once or several times?
(2) It's also good to get feedback when a solution is found, return to the original post to explain how it was resolved so that more people can also use the results.
Dear all, I need your help.
Our team (Department of Construction Management Budapest, Hungary) decided to make a little investigation among softwares that can handle maximal precedence relations. We would like to compare there calculation methods etc,. all in all to perform a comprehensive test of those software that can handle maximal relationships.
(minimal relations from an activity A to B can be the following SSz, FFz, SFz, FSz, where z stands for the minimum time gap betwen the start/finish points of A and B.
Maximal relations define the maximum allowable time gap. (maxSSz, maxFSz,maxSFz max FFz ) with lead/lag where these factors describe the maximum allowable distance
for example B can start right after the finish of B but it has to start within two days.
So far we've got problems in finding commercial softwares that can handle maximal relations. Can you help us to find more?
At the end we will share or experiences with you.
Thanks
Miklos
Printer-friendly version
Send to friend
Carlos,
Sorry for the short disappearing. I had a knee surgery, and there was now chance for using internt in the hospital.
Your latest solution is still not good for me. It is good, that you figured out that a FS1 day with seven working days calendar, can be longer that a maxFS0 with five working days, so solution exist for this problem, and the solution is if A starts on Friday.
But I want you to model this problem on that way that the application has to figue out, that the only solution is when A starts on Friday. In this case you scheduled act. A to Friday and not the application. Am I right?
For example imagine an activity with one day duration for what we need two resources at the same day. We know that one of the resources available at odd, the other at even days only. But we know that there is a certain date (in the resource calendar) when both resources are available at the same day. In this case the software will find the solution and not you. (I hope that almost all the softwares can handle this problem, and give a solution for that)
I want here the same. You define the relations among activities, and the software will find the solution for the problem if solution exists.
Look at my pictures inserted a couple of comments before. A start on Monday and a FS1 (seven working days) relation goes to activity B.
The solution for that is: A starts on Monday, B starts on Wednesday.
In the next step I define a new relation maxFS0 (5 working days calendar) , that is max zero working days can elapse between the start of B and the finish of A. The software found that the first possible solution is when A starts on Friday and B starts on Monday. But this was the application, like in the case of the above mentioned two resources.
In this case you found the solution and you scheduled A to Friday, so you performed the calculation instead of your application.
And at the end if you go back to my ditch shoring example, you will understand that maximal relations are inevitable in a big percent of the projects if you want to make a proper schedule.
Miklos
Miklos,
In the following screencast you can see how I achieved it without using maximal relationships.
I am using two of my favorite features of Spider Project: The ability to add calendars to relationships and the ability to have many calendars.
Basically, I assigned a 7-day calendar to the relationship and a 5 day calendar to the activities. The One-day lag can happen any day in the week but the activities will only be worked between monday and Friday.
I will send you my address in private message for that tokai bottle.
Best regards,
Carlos.
Carlos!
You won the comtest (part a) because you figured out, that if act. A starts (and finish ) on Friday and act. B starts on Monday then both relations will be satisfied. FS1 elpsed day ('causes in this case it's gonna be 2 days) and maxFS0 working day cause it's gonna be exactly like this.
So you deserve it.
However the solution you have provided by the application (part b) is not acceptable. Do not forget: durations must be one day, workers will be payed for one they, so the software has to figure out the right date I have to order the workers (subcontractors) on site. So keep the durations as they were that is one day.
So I need a solution when without maximal relation A would start on Monday and B would start (FS1) Wednesday. You make some modification (but keep the durations as they are) , the software will recognize that there is no solution in this case, (taht if A starts on Monday) and starts to look for a solution which ends with that A starts on Friday and B starts on Monday.
I present a solution in two pictures. First you will se the original situation when only FS1 elepsed day is the relation, then i insert a maxFS0, the software recognizes that maxFS0 is not satisfied, and automatically will starts to look for a solution when both relations will be satisfied. The earliest is when A starts on Friday and B starts on Monday. (I cannot handle this video tool so I'll show xou only the firs picture (without maxFS09 and the last one (with maxFS0).
Without maxFS0
http://www.projackmanager.com/sites/www.projackmanager.com/files/with%20...
With max FS0 (maxFS0 is drawn in blue)
http://www.projackmanager.com/sites/www.projackmanager.com/files/withFS1...
Note that A and B became critical, because this is the only solution within the project duration.
All in all, I do not see if there's a chance to model this without maximal relations, so I'm still looking for the softwares that can handle maximal relations.
Miklos
ps. Do not forget to send me your address to where I can send your first bottle Tocai. Hope security guys won't test it at the airport. :))
Good!
I won a bottle of Tokai! I had never won other contest before :') I am very happy. Send it to Mexico, but I must warn you that I had not to use max relationships (However, you have awarded the bottle so I hope that you will be a gentlemen and keep your word).
I don't know how to upload files. I will send them via e-mail if you wish.
Best regards,
Carlos.
Mistake in the contest example.
Terrible sorry I made a mistake. The right calndars for the relations from A to B are:
B depends on A with FS1 calendar day (elapsed day for MSP users), amd maxFS0 working day.
I have corrected the original text as well.
However Carlos has noticed it and corrected so he gave the rightz solution A will start and finish on Friday, and B will start and finish on next Monday. (Solution in the application will be checked later for the next two bottles of Tocai)
Miklós
Carlos,
Now I think that you really start to understand. But....
You sad:
On Miklo's example, I tied a FF-2 relationship from C to A. In that way I am previewing that A will be finished maximal 2 days before C, and that B will start right after C and when A is ready. But in this case, C did not depend on A.
My comments on this:
1) If you do this - I mean inserting an FF-2 from C to A - it is definietly gives the same result. However we (I) teach that negatíve lags are always the sign of one the follwing three thing
In this case the first option is the right one. There is no logical dependency between C and A. That's a kind of - I do not know how to call it - technical relation in order to express relation between A and B, These kind of solutions always have a major shortcoming, for example they work only in the given case and any alteration in the network will case problems.
To prove this imagine that B apart from A and B depends on twenty other activities, and you do not know, that which one will define the start of B. Just put a new predecesor D for B, where the finish of D is scheduled after C, and you will see that your FF-2 won't work, the distance betwen tha finish of B and the start of A will be greater than 2 days.
I think the best way is to insert a FS0 and a maxFS2 from A to B, because these two describe the real dependency betwen A and B. See my links on my message dated on May 20 19:54 or it is better if I insert them here as well
Without max relations
http://www.projackmanager.com/sites/www.projackmanager.com/files/images/...
With max relations max is in blue
http://www.projackmanager.com/sites/www.projackmanager.com/files/images/...
Contest
On the contest project: Congratulation the right sokution that A can start on Friday and B will start on Monday. So you defietly a winner of the first bottle of Tocai. For the next two I have to understand your demo.
Miklós
Rafael,
Thanks that you still do not give up advising me how to upload pictures. I've tried an another way, I think it works, but I would like to make it better.
If you look one of my first replies my reply #27 (Oh I've just recognized that every time that anyone inserts a new comment, the number will cjhange so I can not quote. Dear admins: could it be numbered on that way, that the first will be #1, the second #2 and so one... so an increasing and not a decreasing way?)
So if you look at my reply (dated on 2011-05-20 19:54) and just click on the link, then you can see the pictures.
I will exchagne all the pictures into URL-s. Thanks again
Miklos
Within step 4 you will have to dissable rich-text to paste correctly the HTML code and then re-set rich text, as this a revised interface of PP and the original posting was prior to the updated web site.
Please use responsibly so we all are allowed to continue using pictures into our postings, it is a privilege.
Miklos,
Now I get the point on maximal relationships clearer. Is it complexity neccessary? We are starting with the paradigm that we must, at any cost, use max rels.
I consider that thinking them as the only solution is too risky, what if there is a better plan?
On Miklo's example, I tied a FF-2 relationship from C to A. In that way I am previewing that A will be finished maximal 2 days before C, and that B will start right after C and when A is ready. But in this case, C did not depend on A.
In the contest example, Activity B is planned to be delayed one day after the finish of activity A, but it must be started maximal with 0 calendar days. This would mean that if Activity A finishes on Friday, it must be started the monday, right? Spider Project handles this easily, look at the video at the end of this page.
Vladimir's example is harder. I started with the assumption that in the best scenario, Activities 3 and 4 will not delay, thus allowing the maximal time (for this example, I am assuming it to be 2 days) between ACT1 and ACT2. But then, what happens in the worst scenario, where ACT 3 and/or ACT4 suffer from delay? I would look for additional activities or resources which could mitigate the risk of delay on ACT3 and ACT4. I can not imagine a maximal relationship if there are not risks involved, in that case I would consider eliminating the dependency from ACT1 to ACT3 and adding buffers: I would plan high productivities on ACT3 and ACT4 to allow them to be finished far before the maximal lag of ACT2.
Here are them, I can provide you with the Spider Project schedule via e-mail.
Best Regards,
Carlos.
To Vladimir and Anon,
Thank you Vladimir,
Your example is a good one the explain maximal relations. Mine was a bit more difficult to understand, because I wanted to present it through a real life example.
I am still working on it how I can insert pictures that will remain for more than two hours, I'll solve it at the wekend, and i'll be back with interesting examples.
Anon
I make itt simpler. Activity B can start right after A and C, (these are FS0 relations), however to protect the result of A, B has to be started within two days after the finish of A, this is a maxFS2 relation. Now imagine that the predecessors of C force C and with this B as well much later then the finish of A. If you do not have maximal relations, then the difference betwen the start of B and the finish of A will be much bigger then two days. With this you cannot preserve the result of A ( eg. the sidewalls will caoolapse, etc.) If you have maximal relaton between B and A, this maximal relation will ensure that the distance never will be greater than two days, on a way that this relation will pull A closer to B, within two days.
I'll put the pictures on, at the weekend. (hopefully in a permanent way)
Until then I want your comments about the following situation.
(A contest for a valuable prize is announced here for all the PP members)
Go ahead guys we 've got 37000 members here, let's fight for this fantastic prices.
Miklos
Miklos,
now I understand clearer your points, and I request you for an excuse because I feel ashamed for saying that these were basic CPM principles. I was assumming that Maximum Relationships would be relationships with lots of lag, which was not correct. it is a shame that your charts are not available now, but after reading the example of the deep hole and after trying to model it, I think that this is a good example os SS + Volume lags relationships, which can only be modeled using Spider Project.
I still don't see the point of "maximal" or "minimal" relationships. The example of the deep hole can be modeled with traditional CPM + Spider's volume lags, you just should define if the TAKE OUT SHORING must be done before or after REFILLING and if the shoring material is available as a unique piece or if it can be placed in "any quantity" immediately after taking "any quantity" of it from another place.
I think that if you make that clear, I can model the work with traditional CPM and volume lags. But if you don't make that clear, I can't get to the point of understanding maximal/minimal relationships.
Best regards,
Carlos.
Hi Vladimir,
This is i want to say and also being doing the same thing from start of planning career.
but i am very much curious to learn and if possible apply what Mr. Mike wants to say.
Hi Mangesh,
SS and FF links with volume lags is the right and the only practical approach in many cases.
Best Regards,
Vladimir
Let's consider an example:
Activity 1 precedes activities 2 and 3, activity 4 follows activity 3 and precedes activity 2.
If dependency 1-2 has maximal limit on lag smaller than activity 3 duration then it will not be satisfied in any case.
This most simple example shows the complexity of applying maximal lag constraints.
Dear Mike,
While planning projects we have use SS anf FF links so as to limit the no of activities. when you say for true critical path only FS links must be used, it will result in no of activities increasing exponentially, which may become unmanageable when only single planner is there on site. in such a case what do you propose, as i would like to implement it when i plan further projects.
Just a query, if we give same Lag for both SS and FF links, can this workout? ( sorry for being novice).
i will study this further by sampling the links on some model projects.
Regards,
mangesh
I am posting my graphs using an external file hosting service. I am uploading my pictures to Photobucket and then I copy the HTML Code after disabling rich-text by clicking on the option at the lower left, then I enable back rich-text to finish editing my posting.
For movies I use You Tube.
http://www.youtube.com/watch?v=VBtl3V1p9HA
In this way free space on Photobucket is almost limitless.
Rafael,
Your charts are still on, mine have disappeared. what's the reason for that?
Miklós
Miklos,
You do not have to convince me traditional CPM logic cannot solve the issue, I am just trying to find a couple of near workarounds.
Regards,
Rafael
Rafael,
The best way of concincing me is two snapshot, one when you show a situation like in your lást post, and the secondm when you change the duration of the predecessor of C in order to force it to right, (this will move the shoring activity to right as well) and as a consequance of it activity A will stay close to C (while a FS0 from A to B must be described as well) I think this can be modelled only with maximal relations as you can see the maximal relation ( plotted in blue) in my previous example. Ups I can not see them, What are the reasons for that?
I include them again
I changed duration of "excavation at hole a" as to switch driving predecessor, instead of "refilling at hole c". I am having some issues with how both models are computing float for some activities, that is why I believe a solution that works within the software can yield better results.
For example in the last model "predecessors to c" can be delayed more than as shown as a result of how the constraint works while float for the start of "shoring of hole a" shall always be 2 days, except when close to overall job finish.
I believe my solution does not answer your question correctly, a workaround but not perfect, just on the safe side.
Maybe initially you shall look for a solution that address the min-max issue between two directly linked activities but not more as this can become exponentially complicated under resource leveling, an issue that shall not be overlooked.
The following figure is from an idea I was developing regarding activities that represent a group.
In this case refilling hole c and start shoring hole a represent a group of activities and perhaps some definition of relative float and strict relationship functionality can be added as an option so you will see float of the group activity but within the group you can see relative float. As whatever I say here is for the benefit of everyone if this can be of help let me know. I had the idea, among others, of making group activities able to work as a Topic Activity in SureTrak, a WBS activity that can be driven by predecessors and that can drive successors. Not liked by the CPM purists, I say burry CPM and extend all functionality beyond, not to be loved by forensic protocols editors that do not understand real life modeling is not that simple and pretend old CPM model can explain everything, especially those who say use of available float is irrelevant and has no impact on the schedule.
Regards,
Rafael
Rafael,
We are getting closer and closer I think :))
Just one more pending issue i do not understand: duration of excavation of hole A seems to be too long. What I need, that if it a two days job that it has to remain a two days job. Shoring shoul pull excavation closer (within two days).
I think the only solution for this is the applicaton of maximal relations or application of no later constraint if instead of calendar date, a relative date could be defined to a start or finish of an another activity. But asi mentioned earlier this is practically a maximal relation
Miklós
You can also apply the constraint at the start node as originally intended, equivalent but perhaps better as it shows true logic.
Dear Rafael
Thanks I'll download it.
Dear Mike,
I agree with you. You do no thave to convince me, we teach the same thing. Smaller sections with FS relation result in better plans. tracking is more adequate and so on.
On the other hand there can be situations where empoying SSz and FFz (z stands for the lag) between two overlapping activity can be adequate, and there is no need to include let's say 5000 instead of 50 activities. For example in the preparation phase or in the early planning phase etc. As planning is an evolutionary process, as we know more and more details we can go deeper and deeper with WBS, we will have the oppurtunity to make the plan as detailed as it can be, when we have the detailed drawings or we have a contractor.
It's the planner responsibility, and he has to decide wisely based on circumstances like time, his own competance, information available etc.
But...I totally agree with you
Miklos
Hi Miklos
I am currently involved in a delay analysis for a 125km gas pipeline.
The contractors progremme showed work in 5 spreads with a single bar for each of the tasks using a SS link with a 5 day lead lag.
There were no resource driven links for different crews or for the HDD rigs.
It is useless for delay analysis so I had to break it up into 1k sections and link it up using FS links
The task took 4 hours using copy / paste on each section.
I was then able to place the crossings and stations into the programme without any guesswork and put the resource links in the correct location.
If the programme had been done like that in the first place there would not have been a forensic delay analysis.
The company are now changing their methods for all future projects.
Best regards
Mike Testro
Here there is a print screen of the logic, it was revised as some links from excavation hole to the dummy was missing, also note on the ALAP constraint on refilling at hole c.
You can download the Spider file at the following link, shall make it easier for your testing.
http://www.mediafire.com/?4sz0192cs41xyjl
Regards,
Rafael
Rafael,
Thanks for your comments. I will download your tool, and will study it. Your notes about loops sound exciting, because when we teach time analysis with maximal relations, we convert the maximal relations into a minimal ones and during this the direction of the arrow (dependency) will change that is a loop wil arise If you considerthe picture in my original post. maxFS2 will change its direction so a loop will arise.
I'll checkMicro Planer as well.
Anyway there are dozens of small and tricky modelling problems that I collected during the years for my students, maybe i'll open a new topic later to share with all of you guys.
All the best
Miklos
Miklos,
I applied the No Latter Than constraint - NLT to activity code 2 by using formulas functionality on the software that allows me to specify formulas per cell instead of per column only, sorry for misspelling the acronym but you got it right.
I used a dummy activity to get the early finish of the "late shoring dummy" and make it the NLT constraint for activity "shoring of hole a". Note there is an inverted yellow triangle that graphically marks on the Gantt the constraint.
I admit it is a way around and for more complex situations it can become very tricky. I believe traditional CPM and PDM must evolve and we must not not be happy with the limited set of modeling rules. If what you propose makes it easier, less dependent on tricks, and less prone to error it should be welcomed. The software shall not be the limiting factor but the end user, lets make it easier for them.
Some people believe traditional CPM rules should not evolve because it is too dangerous to make it easier for the keyboard jockeys. I believe that on the contrary it has become a tool for the keyboard jockey, a tool not used by the people who really understand what is to be modeled because it is too dependent on the tricks such as the one I used and get frustrated with the tool.
Long ago I used Micro Planner for the Mac, I believe it would allow for circular references after issuing a warning they called the "Deadly Embrace", not sure. Was limited as to the amount of activities it could handle but for your research it should be no issue.
There is a free Spider Project demo for unlimited time that allows you to schedule 40 activities, you can download it and if it fits your needs we can discuss whatever on Spider forum in Planning Planet. I have no doubt others will follow with the software of their choice. Maybe our tricks can give you some clues. Remember to give us a few of your own.
Best regards,
Rafael
Rafael,
Thanks, and I think you understood my example well, I've tried to understand your suggestion, but It is not quite clea for mer. I suppose FNL isstands for finish no later constraint, but you assign it to which activity? On the snapshot I still can see the situation when side walls will collpase, The difference between the finish of excavation and the start of shoring is still 10 or more days instead of maximunk two days.
Anyway I do not argue that tricky solutions are exists, while finish no earlier constraints can be replaced by minimal relations for eyample a SF from the project start to the activity, Finish no later constraints can be modlled with a maximal relationship
And especially when Constraints like FNL can be modelled on that way that instead of a date we can give a ralative distance from an another activity, than tis is a perfect solution, But this is norhing else than a relation (max or min depending on the constraint) between two activities.
If the software you use (Spider projects) can handle constraints in this way, than you would be welcome in our test.
For all of experts in this site......We are still looking for softwares that can model not the minimum reqired , but the maximal allowable lag as well. We would like to test them in tricky situations.
Why? Because maximal relations are essential in the vast majority of projects. We made a small research, we have investigated 10 different construction projects with their pkanner, and we were looking for existing dependencies that can be modelled with maximal relationships. They all admitted that the network logic were incomplete,and together we have detected existing aximal relations They also admitted that they have never tried to detect these type of relations because
a) they have never heard of them (8 out of 10)
b) they 've heard of maximal relations but as their software could not handle these relations,
So I'm still waiting for your advise. We have only two application so far and Rafael's if constraint can be defined as a relative to an another activity
Miklos
Got it. Prehaps traditional CPM software cannot model it on a direct way and interpret your requirement as a circular reference if using traditional logic relationships.
In such cases we just add a FNL than constraint, this can be done using formulas as for the constraint date be adjusted by the software using a dummy activity. The following illustration shows a possible way around.
Just one mor notes why do we feel the importancy of testing the calcualtion rules.
In '97 I was writing a book on scheduling. It was a monography not for practioners, but at the end I compared different softwares, all of you know all of them. The original goal of comparing speed, etc. soon be omitted when I've realized, that the time analysis simply comes back with different results for the same network. Couple of problem was detected, just the most interesting.
Consider three activity A, B, C, As these software could handle and still can handle for relations Ss,FF,SF,FF, if one relation is allowed from A to B and one from B to C, 16 different network can be drawn. (No resources, calendar is 7 days 8 hours)
Out of 16 in 12 cases there was no critical path from A to C, and greater than 0 floats were calculated (in case of a widely known application), while in the another applications thete were a critical path in all cases. After a short coresponding they have admitted that they are not right, and they have corrected this in the next version in 98
Very rude problems were detected and there are still some (and in some cases serious) problems in calculations.
So the first thing we teach our students to check the model and the calculation procedure of the applications they use.
Carlos,
Thanks for your comment, I would answer for it, even if it is not connected to my original question, but please be more specific, I could not catch your point.
Mike,
Agree, we teach for our students that the best if you use FS relations, but even if there's an option in your application for using FF, SS, SF you do not have to use them. Here the emphasis was on maximal relations please see my example was sent to Rafael. Only FS relations are used, but in this case lag has to be limited from top as well (with a maximal FS2 relation) and not only from the bottom (FS0 relation).
Anyway in some cases especially when we want to build a draft schedule of a long pipeline or highway FS relations can be used if you divide the whole length into small fractions, in case of a draft SS and FF can be useefull to model overlapping.
Miklós
Rafael,
Thanks for your reply. I'm trying to be clearer with this example. Suppose that a deep hole (A) has to be excavated, and this has to be shored up (B) (against collaption) when excavation is finished.
This is a FS0day from A to B.
However material for shoring will be available after finishing the refilling,(and taking out the shoring material) at an another place. (C)
This is a FS0day from C to B.
This can be done in MS project, Primavera etc. Now imagine that C because of its predecessors will be finished a lot after A, which means that shoring (B) will start a lot after A is finished. This can be seen in picture I. During this time the sidewalls of the hole can collapse. The right logical dependency between A and B would be: we can start with shoring right after the excavation of the hole but must be started within two days in order to prevent collapsing. see picture 2
This is a FS0day and a maxFS2days.
There is nothing about calendars here and nothing about originating about the intermediate point or not. This is about the describe maximum lag between any two activities and not just minimal.
What we are looking for is to collect those softwares can handle these kind of problems and check their calculation in some tricky situations. Maybe not only the introduction of maximal relations are suitable for solving this problem, please advise me if there's any other approach.
without max relation
http://www.projackmanager.com/sites/www.projackmanager.com/files/images/...
With max relation
http://www.projackmanager.com/sites/www.projackmanager.com/files/images/...
Thanks Mikklós
I think that their little investigation should be in scheduling and planning good practices, no offense intended, but what will you do during a greater-than-one-year lag?
Hi Miklos
You are wasting your time on such research.
A true critical path has ONLY FS links - all the others are aberrations.
Best regards
Mike Testro
I believe if using Continuous PDM model there is a possibility some links might be ignored. Therefore the software shall identify links that has been ignored.
I wonder why limiting the relationships to a maximum and a minimum when if using different calendars the driving relationship might be inverted. Why not using more than two similar link types with different lag values and different calendars?
And why ruling out double dependencies where link is not from either end of an activity but from an intermediate point to another intermediate point using independent calendars?
You shall also consider that seasonal or shift work productivity on a single activity might vary, therefore volume of work lag might not be one on one with activity duration. In those cases volume of work lag and duration lag might switch as to which is driving.
Spider Project ignores conflicting multiple outgoing strict links but not sure if identifies them after the schedule run.
Email: PPadmin@planningplanet.com
Replies