Trevor

We do need foolproof solutions, in our country our Project Managers are overloaded and no matter how good they are at CPM scheduling, they simply do not have the time to be playing what if scenarios. In addition not all Project Managers are hooked on CPM, but they are the best qualified to plan the job. Most of our construction jobs cannot afford a full time scheduler on site, but the management issues must be solved as soon as possible, whether there is a scheduler or not, the Project Manager must solve the resource usage issues.

Just read the paper by the CIOB, seems that in the UK the situation is similar to ours and just the opposite to yours. We have no problem at all to be honest with our reality. To fill the gap we need software easy to use and more powerful, that efficiently deal with the issue on resource allocation, otherwise most jobs will continue to be scheduled without the CPM being resource loaded.

http://www.ciob.org.uk/filegrab/TM_report_full_web.pdf?ref=880

Your comment, "I don’t need a foolproof solution because I don’t hire fools to do the job, no matter which tools we use. " is arrogant, sems like you are still taking it personal, there is no need for it.

Best regards,
Rafael

Vladimir,

Can you say the 6,000 tasks and 75 resources is not even a very big problem, but instead of not trivial also trivial to the scheduler when using Spider Project, that Spider Project can get an optimal soulution just at a click of the mouse, without the scheduler even thinking about special settings for the software to solve it? Without need of judgement and care? I am looking for a fool proof solution.

How long it would take for Spider Project to make a single resource leveling run on a notebook computer?

Best regards,
Rafael

Rafael, the rationale is a mixture of judgement, care, careful thought and, yes, trial and error. Trial and error should not be so readily dismissed. It is not a random process, but a methodical and systematic one, usually aimed at eliminating scenarios which don’t work in order to narrow the field and find a "feasible solution space".
Trial and error approaches, such as the branch and bound methods, are well known and used all the time in OR.
When we are modeling a project, which is what project planning is, trying things out helps us to understand the problem. The best thing about the software tools is that the hard slog of the calculations is not an obstacle. Software allows many "what-ifs". Before software there just wasn’t the time available. Even now with software some types of OR problems cannot be solved because it just takes too long.

One trick I find very useful with resource leveling in MSP is to Group By Resource Name.

Vladimir, yes. Bigger just takes a little bit longer.

Mike,

And now you come up with this, after all day taking out all grass from my lawn looking for a square root. As soon as I get a square root I will post the photo. I am already at my neighbors lawn.

Anoon,

Cotto will make a square root out of Pacquiau next Saturday.

Best regards,
Rafael

Trevor,

I am challenging you to see if you give us the rationale, the same way Roland Tannous has provided us with useful guidance in the use of MS Project. I am interested in the logic, not on a single solution.

Maybe Anoon is right and you got the answer (the square root of 25) and then forced the scheduling options to get the answer (5) within MS Project, the easiest alternative would be using the solution by Spider Project.

Maybe you got an answer and we can put to rest all operation research scientists.

If you share your knowledge I will appreciate it, you already know from the spot I accepted your solution to my sample job as optimal and even checked it using SureTrak applying the equivalent priorities values and leveling options and got the same outcome.

If it is so easy then MS Project developers got to be very bad as not to be able to include an optimal solution within their software the same way Spider Project do, about Primavera developers there is no need to say how bad they are, after P6, a disaster from the very beginning, when they started with P3e, and after almost 10 years they are still debugging the software.

Best regards,
Rafael

Rafael, if you think that I was only able able to solve your example problem by inspection and judgement because it was so simple, please provide a more complicated, realistic one and I will solve that too.

I’m for Excel too! you can do what you want (even resource leveling) and never assume any result (whether right or wrong).

i’m never angry

Roland

I will be waiting for your paper, thanks again.

Best regards,
Rafael

Hello all,

Yes I agree that the procedures are pretty annoying.
I guess the ultimate thing is to arrive at a conclusion that there are rules that guide those modifications. IF not we’d be performing some type of hard coding, a clear indication that the software miserably failed.

I am exchanging examples with Vladimir and trying to see if they can be solved in MSP.

When we’re done i’ll try to make a summary and post the examples somewhere and indicate how I arrived to a solution and the reason behind the steps taken. Hopefully that will help all of us understand things a little bit better.
But still i do understand why people get annoyed by it. I am annoyed too.

Regards
Roland

Anoon,

I agree with you, the procedures to get adequate resource leveling in MS Project are insane and error prone.

As of now only Spider Project can do automatically optimal resource leveling after entering restrictions while others do require further cumbersome procedures.

I have to recognize this is the first time I have seen such workarounds (further cumbersome procedures) explained, useful for those of us who have to deal with what we have. Roland presentation has been very enlightening.

Roland procedures can also be applied to SureTrak, what we use at home, although I would prefer Spider Project a lean machine that not only gives you optimal or near optima resource leveling with ease but also provides you with true critical path and float.

To Roland and Vladimir my thanks, I know Trevor will follow.

Best regards,
Rafael

Hi Roland,

I don’t know anything about MSP (or any other program) to be honest, I’m just a part time user in the street corner (no offense, of course).

The first thing you have done is of course MANUAL, and I supposed it applies to all others.

For me, no program is ever intelligent enough to do resource leveling on its own, I mean which can do the settings on its own, and it is all MATH.

Maybe I should ask, how do you compare abacus from a slide rule, or a calculator?

cheers

Hello Vladimir,

The reason why it won’t work if it starts at 8:00 am is due to the working hours for the night shift crew. Go check the Nighshift calendar and you’ll see what i’m talking about.
They don’t work between 8:00 am and 11:00 pm. So if your task starts at 8:00 am and needs a fixed duration of 10 hours to finish, It will finish before its even 11:00 pm. That means before even the Nighshift crew comes to work. That’s why if you assign the Nightshift crew to that particular task you’ll get 0 hours on the Resource usage sheet which makes perfect sense.

check if your activities span a "Monday" . Do your activities span a Monday ?

In addition could you please tell me what exactly in MSP’s leveling output doesn’t make sense so we can discuss it. Maybe I missed something.

@Anoon
No offense, but nothing in the steps I’ve undertaken indicates that I’ve done manual resource leveling.
Most of what I changed are settings in MS Project to tell it that I’m using hours instead of the default "days" value.
I assigned calendars and that’s a step we undertake using any other software too.
Resource Leveling is set to automatic and it did level automatically without intervention..
Regards,
Roland

Hi Ronald,
I’ve got the same poor results with 12:00 as the start time.
If you will get the schedule that makes sense after adjusting resource calendar let me know.
Best Regards,
Vladimir

Hello Vladimir,

I’ve done your little exercise.
I didn’t find anything weird in the leveling exercise in Microsoft Project.
I did though notice that if the settings are not set correctly to start with you will get wrong results and maybe that is what previously happened when you tried to level those resources?
Here is my method and setting:
Step 1) Go to Tools->Options---> Schedule and Change
Change "Duration is entered in " value to "Hours"
Change "Default task type" value to "Fixed Duration"
Step 2) Go to Tools---> Options ---> View
Change "Date Format" to value "days/month/year XX:XX AM/PM"
This step is really crucial because Shifts won’t work with Microsoft Project and it won’t level correctly if you don’t change start time to 12:00AM instead of the default 8:00 AM and if you don’t realise that by default it will assume by default work will start at 8:00 am , its output won’t make any sense. So I change this value to 12:00 AM (00:00 AM)
Step 3) Setup Activities and Enter Durations in hours. Start and End Milestone were used.
Step 4) Go to Project ---> Project Information.
Enter Start date and start time to 12:00 AM
Set Project Calendar to 24Hr
Step 5) Setup the 4 resources as indicated and assigned the appropriate calendars (all use the 24Hr except the one that uses the NightShift)
Step 6) Assign resources to Activities as you indicated.
Step 7) Go to Tools---> Level Resources and
Set "look for overallocations on a ... basis" to "Hour by Hour"


All went well and if look at Resource usage, everything seems fine and the numbers make sense. So I’m not sure what happened with you right there.. Maybe along the way , you missed one of the steps and this gives you wrong answers.
For example Setting the resource Calendars without fixing the Project Calendar appropriately gives wrong results.
Not setting the start time to 12:00 am instead of 8:00 am, which makes sense if we’re trying to work in shifts, also gives wrong results.
Not telling MS Project to look for overallocation on Hour by Hour basis will also give conflicting results

Regards,
Roland

Hello Vladimir,

If a start and finish milestone are introduced, then that’s a totally different subject. For some reason, I had the impression you’re insisting that there would be no links at all, which made no sense to me.

As for "bar charts". Bar charts might be a concept used by scheduling software, but it is a notion that existed before scheduling software did. It is a statistical tool.. People used to draw bar charts by hand. Do not know why people insist that life before scheduling software didn’t exist :)

As for the resource leveling issue. I cannot really talk in the name of scheduling software manufacturers but resource leveling since PMS by IBM depended on some type of Network model or CPM.
They used to go through 3 phases:
1) A CPM nominal time run
2) Resource compilation called unlimited run
3)limitation of resources (constraints) and output.

Even when resource leveling was done by hand, the procedure itself used the concept of critical and float. People who leveled worforce did the following(more or less):
a)draw critical path
b) draw critical workforce
c) plot float paths
d)float path worforce plot.
Level workforce on float paths.

i’ll try to run your example again by using milestones.
As for Rafael’s example, I thought someone already answered him? But I’ll run that too and tell you what i get..

Regards
Roland

Dear Vladimir,

Analyzing and planning workforce and equipment by a network should be preceded with using the basic CPM technique.
Your example obviously fails this simple basic rule, because neither does it constitute a CPM schedule, nor any type of network models. It lacks logical relationships/links!!

From the beginning,Resource planning/leveling techniques relied on the concept of an underlaying Network schedule.
Whether be it PMS by IBM,RAMPS by CEIR, RPSM by Mauchley or the resource leveling algorithms in recent scheduling programs that use the critical path and float concepts to do their job.

So you are expecting people to run a resource leveling algorithm over what really constitutes a bar chart , is not a CPM schedule and the results are supposed to be a healthy indication to whether the resource leveling algorithm is good or not?
For me that doesn’t make any sense.
I mean even if your planning program can do that correctly , this doesn’t mean It is better.
If you really wanna compare try to find an example that is a correct CPM schedule and where some algorithms fail to level correctly then I’m all ears.
But trying to level a simple non CPM bar chart indicates nothing, at least from a mathematical/engineering point of view.
Regards,
Roland

Hello Vladimir,

Define : There are no links?
What is the logical relation between Activity 1 and 2?.

Vladimir, I haven’t (yet) done your example but I think I know what you are getting at. In MSP, if there are two Resources such as a pilot and an aeroplane assigned to various tasks there is no (direct) way to keep them together, so leveling, when done without any care or judgement or attention to the leveling settings, will split them up so that the aeroplane can leave without the pilot. This is well known but it is easy to allow for, such as level only selected tasks or selected resources.

Hey Vladimir,

I don’t think I really follow your logic.
I mean I understand what it is you mean but as Trevor already described, you are able to simulate shifts in Microsoft Project following the procedure he gave you.
The two resources he mentioned using ,lets call them calendars A and B (10 hour shifts calendar) and which derive from a base calendar of 20 hours per day work represent the two different teams which will work different shifts during the day. (The 10 hours in each of resource A and B’s calendars do not interlap, therefore different shifts during the same day).
Apart from that , what is it that you think is still missing. I am not sure I really got your point
Regards,
Roland

Trevor,
all of us have other things to do. So everybody will understand if the reply is delayed.

To similate shift work there is a need to use at least two types of resource assignments:
1) Team assignment - resources belong to one team if they work together and the absence of some team resource means that activity execution will be delayed (machinist will not work without his machine, if the crane is not available the works that depend on lifting something will not be performed),
2) Independent assignments - resources belonging to different teams (different shifts in particular) work independently.

Besides, the duration of effort driven activity shall depend on the efforts of driving resources only.

Best Regards,
Vladimir

Rafael, you said (quote):

"For the moment, I interpret your avoidance on disclosing a methodology as that there is not such a methodology other than trial and error."

I did not avoid "disclosing a methodology" or providing an explanation, I just didn’t do it (yet). Perhaps you should allow for the possibility of other possible interpretations, such as I did not have the time and have other things to do.

We have an American customer - the company that developes the software for the Government. They found us through the Internet.

They had a problem with the Government contract that shall be finished on Summer. Their project consisted of more than 6000 activities that shall be done by 75 programmers with different qualifications. People shall be assigned in accordance with their qualifications and the project levelled. They used MS Project and got poor results despite a lot of hard work. They were desperate.

We imported their project schedule from MS Project, created and assigned resource skills and levelled the project. Spider Project selected who will do what basing on resource skills and the schedule finished on June 14 (starting in November of the previous year) that satisfied the customer.

When this schedule with already assigned resources (not skills) was exported back to MS Project and levelled by this software the resulting schedule finished at the end of October.

Now the task to Trevor:

How to assign activity priorities in 6000 activities schedule with 75 named resources to get satisfactory result?

Manual prioritization means that the planner schedules the project instead of the software. It is easy with 10 activities, hard with 50 activities, impossible with thousands activities.

The schedule that we created for ship building project consisted of 91000 activities. This schedule took into account limited space and some activities were delayed because the place where they can be performed was occupied.
I cannot imagine how to use MS Project for the projects of this size and complexity. Maybe MS Project 2010 will be able to open the projects of this size but previous versions could not.

Trevor, you mentioned that in your project people worked in 2 shifts. But MS Project does not simulate shifts! What workaround did you apply?

Best Regards,
Vladimir

To all,

http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JCE...

"Construction schedules, generated by network scheduling techniques, often cause undesirable resource fluctuations that are impractical, inefficient, and costly to implement on construction sites. This paper presents the development of two innovative resource leveling metrics to directly measure and minimize the negative impact of resource fluctuations on construction productivity and cost."

In the above reference you will find other possible metrics to define optimum solution. Good software should provide you with the best available algorithm for each metric or combination and let you select which metrics you are to use at a click of the mouse.

The knowledge has been there for many years, just that most software developers have not been able to catch up. And yes I believe it to be because of the technical incompetence that we are not getting these functionalities. There are some line coders that can do coding and get nothing good out of it while there are some that can do pretty good things, but these got to be more than just database developers, is more than that.

I don’t have to hide my opinion; this is the spirit of Planning Planet. As long as we continue supporting software that does not satisfy our needs we will be getting just that. The idea of forums like this is not only to trouble shoot specific software malfunction, if so P6 would be the absolute owner of this site. The idea of this forum is also to share knowledge and compare the available solutions one against each other. The idea of this forum is also to expose our needs and make our voice heard, not to be shut-up, make your voice sound loud and clear.

If we want to change the status quo on time management we need efficient resource allocation functionality, easy to use, so that the average scheduler and even the occasional user can make use of the tool easily and without waste of scarce time. If you are an experienced manager scheduling what you know best you should already know the issue is not merely in the logic, the easy part, but also in the efficient use of resources.

Best regards,
Rafael

Trevor,

I don’t believe you have any share with Microsoft; I believe you honestly like the product and know the tricks and traps and one of its fans.

What is the reasoning behind looking optimum or near optimum resource leveling, Brute Force (trial and error)? I would not call “brute force” to be reasoning, not even an algorithm. For the moment, I interpret your avoidance on disclosing a methodology as that there is not such a methodology other than trial and error. Remember the solution worked with SureTrak also, seems like both are using the same outdated resource leveling algorithm, none can find the best solution unassisted. I worked the solution using Spider Project unassisted, this is far superior to trial and error.

Can you please be frank and tell us your reasoning, if there was any other than trial and error? If this is the best Microsoft can do, then my opinion about Microsoft Project will continue being as bad as always. As bad as the opinion I got about Primavera products when I realized how bad at resource leveling they also are.

Finally I would like to say that you should not take it personal by arguing that PP member who have never written a written a line of software in their lives should not evaluate the software they are to use. For years I tried MS Project, my opinion is in no way uninformed, is based on comparison with the functionalities of other available software. I can even remember MS Project version for the MAC, this was really bad compared to Micro Planner version available at that time.

Best regards,
Rafael

Trevor,

Thanks for standing up to the challenge. Optimal solution shall yield shortest project duration feasible within resource constraints. Preferably the software should look for it without requiring your intervention as I did with Spider Project, no small reasoning required, none at all. The software has optimization with the only available input for priority as optional, for my run with Spider Project I did not used priorities.

With SureTrak default settings I got 87 calendar days duration after resource leveling, a 21% difference compared to the 72 calendar days.

I applied the priorities and leveling settings to SureTrak and to MS Project as per your recommendation and it worked on both as you just said, got 72 calendar days duration after resource leveling

Please explain your reasoning so I can apply it to other jobs with hundreds of activities, I will appreciate your sharing of a few tricks (all) on resource leveling.

- "Why would you have ever believed that the software can provide an optimal solution?" : I assumed they were using some operations research techniques, to me any solution not close to optimal is no solution. The same goes for manually tricking the software by trial and error to produce a better solution.

Of course trial and error by the computer without external intervention is acceptable. A logical procedure to select your leveling options that will ensure optimal or near optimal solution is what I would consider slightly better than nothing, but no less, still close to nothing to what today’s standards should be.

Best regards,
Rafael

Hi Toby

I look forward to receiving my copy for peer review.

Best regards

Mike Testro

Toby,

I believe more emphasis should be given to resource planning. Ups and downs in resource demand create inefficiencies at the jobsite; it can even influence the perception of urgency as demand goes up and down.

Many of today’s planners do not consider resources loading into their schedules and to make it worse, those who do are using the wrong software. All Primavera Products, Microsoft Project and almost all do provide substandard resource leveling using outdated algorithms to handle the issue. They still use 40 years old and outdated reasoning.

To my surprise a few weeks ago I realized most software resource leveling is not looking for optimization but for a less than optimal solution. Naively I assumed they were looking for the optimal or near optimal solution. These produce resource leveling plans that misguide the Project Manager.

The good thing is that you can test the resource leveling algorithms with real life sample jobs without the need to understand operations research science.

Try resource leveling with the following job to see how your software can handle it and let me know. All relationships FS and 0 lag, two resources AA and BB one each assigned per activity as per table, one of each are available, for all activitiees use a 5 work days per week calendar, very simple.



Answer: Before resource leveling 70 work days total job duration, after resource leveling 72 work days days total job duration.

If your software cannot solve it then try to use “soft dependencies" to solve it manually by your own, got to be easy, only ten activities. of which two are milestones.

Best regards,
Rafael

Trevor,
you suggested me to pay for the decision I don’t need.
You asked for the real project, not me.

Spider Project finds the shortest resource constrained schedule for this project in 8 seconds. And my notebook is two years old and is not the fastest machine available.
You suggested to pay for the long time you will spend finding similar solution and honestly I don’t believe that you will ever do it.
This way you answered the question why the software that creates good schedules is profitable to have. It will save a lot of time, a lot of efforts, and a lot of money.

And having good and experienced field managers does not mean that the scheduling is not necessary as one can decide reading your post.

Best Regards,
Vladimir

Hi Toby

Thank you for that head up.

I will contact Gary at PEO with a view to getting involved.

The last I heard of Keith he was in Hong Kong but it will be good to meet him again.

Best regards

Mike Testro

Hi Toby.

Who is going to do the training?

It is hardly touched upon in colleges and universities so no one knows the basic principles when they start work.

How do you stop contractor’s planners putting together multi task multi location bars and then using mid bar links to the next bar.

In the analysis I am now working on the planner has one bar for "1st Fix All Trades" over a three storey building linked to "Plastering" and then on to "2nd Fix All Trades" each with a start-start link with a 5 day lag.

This proctice is not unusual - in fact it is alarmingly common.

I welcome your suggestions as to how to get the next generation of planners to do it properly - the current lot are beyond redemption.

Best regards

Mike Testro.

Hi Ken

All programming software allows you to over ride logic links and force a task to either:

1 Start on a set date
and/or
2 Finish on a set date

These are what I call artificial restraints - they distort the logic and the critical path and therefore should not be used.

If there is a reason why a task cannot start before a particualr date then put in a predecessor bar and give it a name.

Best regards

Mike Testro

excellento

will check it out :-)

thanks Mike,

very informative. Guess we arent doing too much wrong (we do love those lag links ha ha). I totally hate any artifical constrainst. As far as I am concerned, people using those shouldnt be called planners!!!!

Tell me, what is the joy of using powerproject compared to other software? It is not used down here in sunny south africa.......

Hi Mike (again)

Please can you explain the requirements that make a contractors programme fit for purpose (so to speak). I am always wondering about the balance between too much and too little information.

Thanks

Danya

Dear Toby,

Since you are talking about complex projects, then the data required to plan and schedule these projects are always available. You will have to depend on the project team and their experience.

Training is always good and it will enhance your skills only if you are using what you trained for. With complex projects comes, complex requirements as well.

The key to success in these projects would be planning, execution and control and monitoring the activities on a daily basis. If you have a good team and enough technical support and resources, you will be able to complete your complex project.

If you have something specific, please let us know.

Best Regards,

Samer

Anoon,

By taking a look at the Forum Discussion Category Primavera Version PM5 / PM6 you will see how many questions are there regarding the many bugs within the software.

Is even hard to get simple graphic reports with functionality similar to old P3, is more like you are limited to screen or layout printouts. This was an issue about 10 years ago when P3e was introduced, a software different to P3 and developed by another enterprise who sold the software to Primavera.

Primavera P6 has so many date definitions it even creates confusion, and some data fields seems like temporary calculation values.

Primavera P6 does not gives you a single automatic algorithm that look for optimum resource leveling, it cannot even provide you with true float values under resource leveling. Of course the various definition of optimal solution might require various algorithms. My main concern is job duration, for others might be resource smoothing, not as defined by Primavera but to minimize incremental changes in resource loading, for others might be Time-Cost trade off to minimize total project cost as a function of project duration.

The software is pricey and you can get better value from other software that is more efficient and less demanding on the hardware. At home we consider it too expensive and we have serious issues with the Government agencies, mainly uncle Sam, the US Government, because even against the federal procurement rules (FAR) many US Government Agencies brand specify Primavera, you have no option, is a matter of force, bullies take advantage of it.

What can you do when your own institutions evade the law, as the procurement regulations are supposed to have the full force of the law? I guess is complain, and make your voice heard. If not because of this I would not be comparing other products with Primavera, there would be no need. I am for allowing the end user to select the tool. If you are working on a lump sum job then the Contractor should be free to select the tool of his choosing, if you are under a PM type of agreement then the PM is the one who shall select the tool. We are in need of government agencies to stop evading the law. We need to get our government agencies, State and Federal, to issue performance specifications that do not favor a particular software.

GSA serves as the acquisition and procurement arm of the federal government, the following was extracted from one of their job specs.



Obviously, under free choice, I would expect MS Project to be the most widely used in Government Contracts, and why not?

Best regards,
Rafael

Roland,
those who use Spider Project consider MS Project interface as not logical and hard to use.
It is natural.
Besides, Spider Project has much more functions and thus options. It makes its interface more complicated.
We will work to create an option to hide the functions that are not necessary now.

I know about some of MS Project scheduling add-ins. Some of them produce better schedules for small projects but work poor with the large schedules. But I don’t know all of them. So please give us the reference.

Best Regards,
Vladimir

A question for Mr. Rafael Davila (if I maybe allowed):

What really made you think and said that P6 is the worst?

Is it in general? If not, in terms of what?

Thanks