Resource Overloads Bar and Resource Dependencies

Anoon,

I make conclusions about your schedules basing on your posts. And these conclusions are not on the quality of your schedules but on the projects you deal with.

The schedulers do what they are told to do (if they agree with this). So the problem is in the poor understanding of the value of proper scheduling by most top managers. In the companies where top managers understand that proper scheduling can save time and money the scheduling itself is the team work and planner/scheduler participates in decision making becoming very important member of the management team.

Anoon,

I make conclusions about your schedules basing on your posts. And these conclusions are not on the quality of your schedules but on the projects you deal with.

The schedulers do what they are told to do (if they agree with this). So the problem is in the poor understanding of the value of proper scheduling by most top managers. In the companies where top managers understand that proper scheduling can save time and money the scheduling itself is the team work and planner/scheduler participates in decision making becoming very important member of the management team.

Vladimir, Your experience or my experience is not the issue. In fact I'm even considering you as a mentor (only here in PP), but I'm sure that you have never seen a schedule that I personally did. To be honest, being just a mere employee as a scheduler (like me) for example, at least from my own experience, you cannot do anything but just to follow the Bosses. That's why I referred to so-called "best practices in scheduling", I mean in general (not you). Which for me, never really happens in real-world. And I suspect, simply because the importance of "Planning and Scheduling", perhaps is not really understood in the industry. So who would care about a certain software's output (leveling or whatever) for example? And most project managers don't even know how to use a "scheduling software". Believe me they're just happy with an Excel spreadsheet. And of course when you go to Engineering and Procurement, they are all busy doing their own things. So what do you think a "poor scheduler" can do? Just look at some of the posts here in PP. Most were asking for a sample schedule. And if you give them the sample, do you think it will help? For sure it can make their schedules even worse. As I've said, while Engineering has a properly defined process or flow of works, Scheduling has nothing, but must cover all disciplines including Procurement etc. That's how difficult "Planning and Scheduling" is! So "Schedulers" raise your Rates, or just let those companies go bankrupt! Where's the "Best Practice" by the way?

Anoon,

I did not talk about best practices.

We have different experience. Your schedules never include parallel activities competing for the same scarce resources, my schedules always meet this. That's all.

As I always wrote good schedule model shall include everything that shall be done (quality assurance activities, commissioning activities, as anything else) and risk events that may happen or not.

Vladimir, Yes it is a fact that there are no "FS only" schedules, as of course you can always do multi-tasking (again) depending on availability of resources, but if you got only one resource, then the only solution is FS only. Even computers or software will tell you that (as computers never lie). However, computers or software (likewise) cannot tell you the truth, especially in terms of "out of sequence". Perhaps the problem is always "misconception". As I guess, "planning & scheduling" as a job is always underestimated or nobody thought that "planning & scheduling must cover all disciplines (QA/QC and Safety included), and never specific only to "time scheduling". Perhaps I may suggest that "Planning & Scheduling (like Engineering) should be done, or be considered as a process, because you can never do proper scheduling without any basis. For me, Scheduling has no difference in "Design or Engineering" as a job. While Engineering has FEED, Basic, Detailed or whatever. You can only find "Schedulers" always busy browsing the web (or Planning Planet in particular), trying to find a source where to copy a certain schedule. So apologies, but what "Best Practices" are you talking about?

Anoon,

of course the schedule shall consider everything that is required for sucessful project delivery.

It does not mean that there is only one way of activity sequencing with FS links only.

Vladimir, Of course there are many ways to execute a certain job. Like "design" or engineering for example, but as you know, even the best engineered product sometimes fail. For me, "construction scheduling" is always "engineering based", where ideally, you should consider the approved method statement, quality control procedures and quality assurance, and of course available resources in order to derive a good project model. Otherwise, for sure your schedule is wrong. Hence, you cannot simply base construction scheduling in the shortest possible time or duration, as calculated by the software. As software itself simply cannot determine "out of sequence". Plans remain as plans until realized. And there should be no shortcuts when you are working with quality control and assurance.

OK Anoon, I just see that you don't want to answer to my question.

It looks like you never met the situation when the work can be done different ways. I've met with this a lot of times. And I agree that when the work can be done in only one straightforward sequence the scheduling and leveling (and any scheduling software) are not needed and out of sequence may never happen.

Vladimir, Sorry but I don't see any logic in your sample that perhaps made you believe that when using the same resource for similar but another activity, both activities can possibly be done in parallel (or start to start). I guess this is the effect of using "unnamed samples" everytime. Of course you can be blinded by math or theory. Perhaps it is forgotten that activities and resources must "jive". In other words, a dancer can only dance on a beat at a time to maintain the rhythm. If you play two different music at the same time, do you think a dancer can dance? Possibly, but of course there is no rhythm. And that's how I see your sample.

Anoon,

in my example activities 1, 2 and 3,4 do not depend on each other. FS are linkes between activities 1 and 2, and between 3 and 4.

And creating artificial links lead to poor schedules like the order that you suggested. Doing all activities one after another though they may be done in parallel is poor solution.

I think that my question was clear and that you can suggest the best schedule of this 4 activities project.

My question was if your decision on the order of activity execution in your schedule was based on activity floats.

Good project model includes only those activities, dependencies, resources, etc. that are real. The software is used for finding the best way (best order) of activity execution. If the model is good and includes all constraints that people consider when schedule their projects manually the software provides feasible results.

Adding any artificial constraints and dependencies restrict software options for finding the best solution and lead to worse schedules. But sometimes preferential logic may be added (to build the road from point A to point B or vice cersa?) and in this case different options shall be tried for finding the best solutin.

So the model where the order of work was already selected by people is not a model at all. It is the result of manual leveling that is poor in most cases.

In the second link there is a column called Preceding Activities, you will see that all links are FS.

Bogdan, I'm not able to see the links, but yes I agree with only one thing: "Using A Good Project Model". I'm not sure about "Resource leveling Algorithms". I can only guess that if you got a "good project model", then all algorithms shall follow, and perhaps you don't even need the so-called resource leveling (as your assumptions are already good in the first place).

Vladimir, Anoon,

If I may, here is a basic model I made for a more complex project, with the help of Vladimir. The main purpose of the basic model was to find a way to solve the following problems:

1. Project must finish in a maximum of 270 day;

2. Each street must finish within 60 days;

3. No overallocations.

All links are FS.

The following capture shows the activities of 1 street:

https://i.postimg.cc/MZnnSBnT/Street-Activities.png

The next capture show that for 1 street there are no overallocations:

https://i.postimg.cc/CMRg1HC3/One-Street.png

Next is the project scheduled with CPM and has a lot of overallocations:

https://i.postimg.cc/rmdP8g6g/No-Leveling.png

Next is the solution provided using only the well constructed model and resource leveling:

https://i.postimg.cc/05wBFt0y/Leveled.png

Now Anoon, the sample consists of 9 streets with no overallocations/street, but the real project has 30 streets with overallocations, how many iterations do you believe a scheduler would have to make until a right solution would be found? How many if each street would have overallocations?

Using a good project model and a configurable resource leveling algorithm it would take you minutes instead of hours or days.

Best regards,

Bogdan

Vladimir, Then the sample is wrong. If you have used FS only, your schedule would be: 1 -> 2 -> 3 -> 4. There's no way that 1 & 3; and 2 & 4 can have start to start relationship (even if using ASAP). That's why I said the sample doesn't make sense.

OK, Anoon,

just imagine that resource A in my example is fitter and resource B is welder.

Activities 1 and 3 require fitter and activities 2 and 4 need welder.

In CPM schedule activities 1 and 3 start simultaneously that is not possible since you have only one fitter (or fitting crew) and only one welder (or welding crew). So it is necessary to decide if to start with activity 1 or activity 3.

I asked about your decision and if it is based on activity floats.

You suggest to rely on scheduler intuition and manual leveling. I don't think that the scheduler can manually level large schedule and get the best possible solution and then repeat this process when any information was changed (activity delay, changes in resource availability, changes in productivity estimates, etc.). Manual leveling is hard and time consuming job. Our schedules consist of many thousands activities and manual leveling is not an option.

Vladimir, I'm not sure if what I'm seeing in my very small screen is what you have really posted. Of course when you run the schedule, considering all FS, then it will just follow logic. Say resource A is overloaded (but I cannot see it in the sample), then you run resource leveling, but if you're using FS, I cannot see a reason why it will be overloaded. Anyway, forget about the sample (honestly, it doesn't make sense to me). Just consider two resources for example: Pipe Fitter and Welder (just consider them separate, but I know most pipe fitters are best welders as well). Of course, these types of resources can only be sequenced in FS (you cannot weld until the pipes were properly fit). If you use FS only, there would be no overloading, unless there are other WBS occurring in parallel and were assigned with exactly the same crew or resources. If this case happens, then the scheduler must decide if which WBS to be prioritized or which one to be delayed. Of course the software can never do it by itself.

Anoon,

let's look at the sample project below

Resource A is overloaded and so it is necessary to decide what activity to do and what to delay?

All dependencies are FS.

What is your decision and is it based on activity float?

Vladimir, I didn't get it when you said "there are other factors considered other that floats". Aside from scheduling ASAP or ALAP (Early or Late), then what is it? The only thing I can guess is to adjust OD or original duration by shortening or lengthening it. Let me go back to the basics: A project (or schedule) is simply defined as:"Series of related activities with a beginning and an end"(period). Of course it does not mention floats, logics, resources or even duration, but as you know, a schedule is even more than, floats, logics, resources or durations. I would like to define the project or schedule as: Series of related activities with planned resources and durations (period). I missed the proponent of "DRAG". What happened to "DRAG" by the way? (anyway, I still don't understand it).

Anoon,

as I wrote earlier activity floats is only one of many factors that are considered in resource constrained scheduling.

The aim - to create the schedule that has minimum overall duration taking into account all existing constraints. The software will not prefer one schedule to another if these schedules have the same duration.

I never suggested to use retained logic only because people may miss something when schedule model was created or something may change. Out of sequence may happen if the schedule model is not perfect and shall be adjusted.

Resource overloads happen when the schedule was not properly leveled.

Vladimir, Sorry for the multiple posts (I hope Mike T. will delete the duplicates). "Work Order" or sequencing always matter in construction scheduling (even without considering CPM). I guess today's software are all CPM based, and the so called "AI" or perhaps algorithm simply based its so-called decision only in terms of activity "floats" (nothing else). If the schedule is out of sequence in the first place, of course the software can never tell that (only humans). Therefore, the "out-of-sequence issue" alone cannot be resolved by any software, but perhaps shall aggravate it. Thanks for the option "Progress Override" (the issue can always be resolved by smart Humans). For me, there is no such thing as "Resource Overloads", but only wrong assumptions in the first place (or wrong scheduling or sequencing or wrong resource assignments). It is a fact that resources in construction are always limited (or there is no such thing as unlimited resources in construction. Otherwise, contractors can never go bankrupt). Hence, "unlimited resources" or "any order or sequence" should never be an option in "construction scheduling". Then perhaps you can insist on "Retained Logic Only" as for sure there would be no "resource overloads" or resource leveling necessary.

Vladimir, Sorry for the multiple posts (I hope Mike T. will delete the duplicates). "Work Order" or sequencing always matter in construction scheduling (even without considering CPM). I guess today's software are all CPM based, and the so called "AI" or perhaps algorithm simply based its so-called decision only in terms of activity "floats" (nothing else). If the schedule is out of sequence in the first place, of course the software can never tell that (only humans). Therefore, the "out-of-sequence issue" alone cannot be resolved by any software, but perhaps shall aggravate it. Thanks for the option "Progress Override" (the issue can always be resolved by smart Humans). For me, there is no such thing as "Resource Overloads", but only wrong assumptions in the first place (or wrong scheduling or sequencing or wrong resource assignments). It is a fact that resources in construction are always limited (or there is no such thing as unlimited resources in construction. Otherwise, contractors can never go bankrupt). Hence, "unlimited resources" or "any order or sequence" should never be an option in "construction scheduling". Then perhaps you can insist on "Retained Logic Only" as for sure there would be no "resource overloads" or resource leveling necessary.

Anoon,

if the order of work does not matter (project duration is the same) the software may select any order.

If there are some preferences then enter activity priorities and the software will select the preferable order.

Vladimir, I'm not arguing about the importance of computers or software. Of course the "right software" is a necessity in todays construction. However, for decision making, I believe that you cannot just simply rely on a software output (regardless how reliable a software is - or simply GIGO). Let me cite real examples as follows: You got a pipeline project where you were required to use three types of pipes as: RC Pipes, Steel Pipes, and FRP or GRP Pipes. And you got one specific crew for each type of pipe. However, you had only one "special crew" in-charge of Pipe Fitting which can be used or common to all pipes as mentioned (just for example). Say you have started the three segments (three types of pipes) at the same time and assigned with each corresponding crew. And the "one special crew" was assigned to all the segments at the same time as well. Considering all the segments (all types of pipes), were equally critical as scheduled, how can the software select which segment would be prioritize (or delayed) with respect to the "special crew" assignment? Considering no "human decision" has been made yet (or you just let the software decide first).

Vladimir, I'm not arguing about the importance of computers or software. Of course the "right software" is a necessity in todays construction. However, for decision making, I believe that you cannot just simply rely on a software output (regardless how reliable a software is - or simply GIGO). Let me cite real examples as follows: You got a pipeline project where you were required to use three types of pipes as: RC Pipes, Steel Pipes, and FRP or GRP Pipes. And you got one specific crew for each type of pipe. However, you had only one "special crew" in-charge of Pipe Fitting which can be used or common to all pipes as mentioned (just for example). Say you have started the three segments (three types of pipes) at the same time and assigned with each corresponding crew. And the "one special crew" was assigned to all the segments at the same time as well. Considering all the segments (all types of pipes), were equally critical as scheduled, how can the software select which segment would be prioritize (or delayed) with respect to the "special crew" assignment? Considering no "human decision" has been made yet (or you just let the software decide first).

Vladimir, I'm not arguing about the importance of computers or software. Of course the "right software" is a necessity in todays construction. However, for decision making, I believe that you cannot just simply rely on a software output (regardless how reliable a software is - or simply GIGO). Let me cite real examples as follows: You got a pipeline project where you were required to use three types of pipes as: RC Pipes, Steel Pipes, and FRP or GRP Pipes. And you got one specific crew for each type of pipe. However, you had only one "special crew" in-charge of Pipe Fitting which can be used or common to all pipes as mentioned (just for example). Say you have started the three segments (three types of pipes) at the same time and assigned with each corresponding crew. And the "one special crew" was assigned to all the segments at the same time as well. Considering all the segments (all types of pipes), were equally critical as scheduled, how can the software select which segment would be prioritize (or delayed) with respect to the "special crew" assignment? Considering no "human decision" has been made yet (or you just let the software decide first).

Vladimir, I'm not arguing about the importance of computers or software. Of course the "right software" is a necessity in todays construction. However, for decision making, I believe that you cannot just simply rely on a software output (regardless how reliable a software is - or simply GIGO). Let me cite real examples as follows: You got a pipeline project where you were required to use three types of pipes as: RC Pipes, Steel Pipes, and FRP or GRP Pipes. And you got one specific crew for each type of pipe. However, you had only one "special crew" in-charge of Pipe Fitting which can be used or common to all pipes as mentioned (just for example). Say you have started the three segments (three types of pipes) at the same time and assigned with each corresponding crew. And the "one special crew" was assigned to all the segments at the same time as well. Considering all the segments (all types of pipes), were equally critical as scheduled, how can the software select which segment would be prioritize (or delayed) with respect to the "special crew" assignment? Considering no "human decision" has been made yet (or you just let the software decide first).

Vladimir, I'm not arguing about the importance of computers or software. Of course the "right software" is a necessity in todays construction. However, for decision making, I believe that you cannot just simply rely on a software output (regardless how reliable a software is - or simply GIGO). Let me cite real examples as follows: You got a pipeline project where you were required to use three types of pipes as: RC Pipes, Steel Pipes, and FRP or GRP Pipes. And you got one specific crew for each type of pipe. However, you had only one "special crew" in-charge of Pipe Fitting which can be used or common to all pipes as mentioned (just for example). Say you have started the three segments (three types of pipes) at the same time and assigned with each corresponding crew. And the "one special crew" was assigned to all the segments at the same time as well. Considering all the segments (all types of pipes), were equally critical as scheduled, how can the software select which segment would be prioritize (or delayed) with respect to the "special crew" assignment? Considering no "human decision" has been made yet (or you just let the software decide first).

Anoon,

when several activities require the same resource at the same time it is necessary to decide what activities to delay. Wrong decision may lead to the schedules that are longer than necessary. That is why the software looks for the decisions that minimize overall project duration. Some algorithms are based on activity floats but it is not sufficient for finding the best solution. Spider Project uses sophisticated optimization algorithms that usually create shorter schedules than created for the same projects by other software. Look at this discussion http://www.planningplanet.com/forums/planning-scheduling-programming-di…

So right decisions shall be based on the consequences to overall project duration and even if you assign scarce resources to activities with largest float it does not guarantee the best schedule. It is necessary to look much further than to activities competing for the same resources at current moment. That is why we need computers - to help us to take right decisions and not only to print nice reports.

Vladimir,

My understanding of "resource leveling" is simply this (please correct me if I'm wrong): Perhaps there are several activities within the schedule that were assigned with exactly the same crew or resources and happened to occur or were scheduled to be done at the same time. In other words, there are activities in the schedule competing or requiring exactly the same resources at the same time. Now, what you gonna do if you got only one crew and there are two or more activities needing the crew at the same time? Of course "human decision" would be necessary, if which activity would be prioritized. Otherwise, the software I guess will simply select the activity randomly or perhaps based on floats. And of course the schedule can possibly turned out to be physically or literally out of sequence, and the software of course do not know that.

Anoon,

it is obvious that it is not possible to show real life schedules in the forum discussions. So discussing scheduling problems people use examples.

Now imagine that you got following real life task:

Determine what numbers of people (of different specialities), machines and equipment you will need at the remote sites and what spreads to create for building 730 km pipeline to certain date and with other time constraints like the need to build certin parts of it at predefined small time windows?

Using more resources than necessary means additional expenses, using less resources than necessary means finish delay and potential penalties. Missing time windows mean 6 months delays.

We did this creating project model consisting of 9000 activities and optimized resource requirements using resource leveling and risk simulation. Can you suggest any other way? I don't think that superintendents look for many months ahead deciding who shall do what and in what order.

As you certainly know resources are always restricted. And the software can estimate much more options for finding the best way to do the job. But it may be useful only if it takes into account all existing constraints like do the people that make decisions. Superintendents will rely on the schedules created by the software if it takes into account everything they do. 

We met disbelief on construction sites many times but later, after system implementation, the same people told us that now they don't understand how did they work earlier.

In my humble opinion, to use "ALAP" in a schedule, first you need to decide. Which perhaps brings me to conclude that: "Resource Constrained Scheduling" cannot be done automatically or you cannot just rely on the software to do the "resource leveling". Most samples or presentations that I've seen regarding "resource leveling" only uses unnamed activities and resources which never represent specific or realistic activities, but simply projecting CPM theory (again and again). So I guess people in the fields will never understand the theory of "software generated resource leveling". A construction superintendent for example has two teams or crews to do the pipeline, which he can able to place each crew in any segment along the pipeline, simply based on his own personal decision. Do you think the superintendent needs the software to determine exactly where to place his crews in order to optimize the time schedule?

Anoon,

ALAP (zero free float) is used when you need something to be done "Just in time".

Example: it does not make sense to deliver materials long before they are required.

Negative float is an information that helps to analyze project status but is useless for project network analysis. Even if my project is late I still need to know schedule critical path and activity floats in the current schedule calculated creating forward and backward paths from achievable dates. By Critical Path I mean Resource Critical Path, traditional Critical Path is its particular case if project resources are not limited.

Vladimir, Normally and firstly, you always create schedules using "ASAP", of course to have buffers or floats. And the last thing you want is to run your schedule on "ALAP" and consume all your floats. You never do ALAP unless there is valid reason (say your resources will not be able to start ASAP or as early scheduled, but perhaps within the "free float" or as late as possible or ALAP, which simply means zero float). And creating or projecting "resource assignment" is somehow a fact, which for me is the "valid reason", but not "ALAP perse". So for me, "reasons of delays" should always be based on facts. As regards "negative floats", these will only appear if your project (or certain activities), completion dates were constrained. For me, "negative floats" are not bad, but of course "constraints" are really bad in scheduling. However, in real world scheduling, you can never avoid constraints, "contract completion dates" for example. Negative float will show the exact amount of delay in terms of time, and of course it is a warning, as negative float simply means delay. Now, what I'm trying to say? Who's saying that "Resource-Constrained Scheduling" is easy?

Anoon,

looking at the Gantt Chart you may notice that some activities do not start at the early start day and may be interested why. One of the potential reason - an activity start was set as ALAP, other reasons may include the lack of required resources, materials, costs. This is what we call Delay. It does not consider due dates and so negative float does not matter.

Vladimir, No, the activity is not delayed even if not able to start on "early date", because you have ALAP. If you have started beyond ALAP, then surely you are delayed and of course you will have "negative floats". By the way, the very first time you create a schedule, of course there must be no "negative floats" (nobody wants a late schedule). Negative floats are only based on facts (or it will only exist once you start updating your schedule). That's why I would like to reiterate what I said earlier: Schedules perse cannot depict the reasons of delay (until you have facts). So again, forget about using "retained logic only".:-)

Anoon,

if an activity does not start at its early start time then it is delayed for some reason. Activity delay does not necessarily means project delay. And negative float also does not mean that an activity was delayed. Maybe target date was too close or some of the previous activities took longer than expected.

I don't like and approve negative floats. Floats are used for schedule analysis and if an activity is critical in the current schedule it has zero float even if the project is late or some time constraints were not met. Spider Project calculates (but not uses) negative floats also (as a separate field) but just for the reference and for people who are used to this.

Activity resource constrained floats and resource assignment floats are not the same. Look at slides 56 and 57 of my presentation at recent Construction CPM Conference:

http://www.spiderproject.com/images/img/pdf/Construction%20CPM%20Confer…

If there is float, then it means there is no delay. ALAP simply means zero float, but still no delay. Hence, "reason of delay" is blank. I guess a schedule perse cannot depict "reasons of delay" as "delays" are based on facts. Otherwise, if you have "negative floats", then the "reasons of delays" would be obvious (activities with negative floats). Now, this is the problem: How do you determine resources with negative floats? Or how do you calculate floats for resources?

Anoon,

I don't understand what are you trying to prove.

Of course there are uncertainties and of course they shall be taken into account when contingency reserves are calculated.

Nobody here suggest to use "Retained Logic Only".

But life is more complex than you proposed, FS only does not work in most projects and particular in the pipeline construction projects that you used as an example.

LOL...simply "free float" and yes ALAP. The problem with samples is that you don't know exactly what you were trying to represent (but CPM theory). Try using real example for piping: Cut; Fit; Weld (FS only). Say Act. 2 is Fit. Then you will realize that your example is not applicable in real world.

Vladimir, Everything in real construction is uncertain. Nobody knows when a riser pipe for "pumpcrete" will be clogged or the tower crane will breakdown (during concrete pouring for example); or condemning a just recently poured bored pile (because of inconsistent concrete pouring and suspecting cavities). And of course some original logics will be violated, depending on real situation on site. If you intend to pour a slab monolithically for example, then if anything happens during concrete pouring, then you must be prepared to provide for correct construction joints which may need application of epoxy or water-stop or whatever for example. And of course these are all violations of original logics. @ Rafael: "What really drives schedule logics"? It is not your plan or schedule, but real site conditions and of course people on the grounds. So forget about using "retained logic only".

Rafael,

Considering what you said, I must agree with you, "calendars" are natural object atributes, and shuold not be considered a delay generator during the natural scheduling process, in the same way we could say that the ALAP modeling condition would also be a planning attribute, not a top reason of a delay.

Anyway, the "report of the reason of delays", is really very useful for planners.

Anoon,

adjustments do not always mean violation of approved logic.

You described uncertainty that shall be considered in risk analysis.

In any case, inconsistent productivity rates can be a cause of delay. Even truck mixers stuck in traffic can cause major delays (just imagine when you had just started pouring bored piles and succeeding mixers were stuck in traffic). These delays cannot be seen in schedules at first, but when you update your schedule, you need to reflect what actually happened. So again and again your schedule always need adjustments. That's why "retained logic only" is never a "best practice".

Hi Rafael,

could we include, "not 24hd7dw Calendars"?

What can be a “Reason of delay”?

1. Actual Start.
2. Active Driving link(s) [can be more than one].
3. Active Resource dependency(s) [can be more than one].
4. Active Start [NET] Date Constraint.
5. Data Date, activities not started with no active driving links, no active driving resources and no active driving Date Constraints will be placed at data date.
6. Concurrent delay reasons might occur.
7. Not "24hd7dw Calendars" applied to resources or activities.