文字起こし 🔗
Bahadir Barbarosoglu: agree.
MIchael Bograd: welcome back. My name is Mike Bogret. I am a vice president at Stellic. With me today we have Bahadir from Holder Construction. Bahadir, we've worked together for a long time now, we've known each other for a long time. Why don't you you know give everyone the elevator pitch of of Bahadir? You know, who are ya, what's your experience been in, what is your project controls experience especially?
Bahadir Barbarosoglu: So thank you very much for invite, Michael. Like I said, my name is Bahadir. I go by Baha. I am a civil engineer by background, bachelor's, master's, and PhD in the same subject. I have been in the construction industry for about 15 years now. The last eight years has been in the data center industry. I have been a consultant to general contractors. I have been a consultant to owners. And finally, I am a general contractor. I'm a director of project controls with Holder. I joined Holder about, you over two years ago now.
MIchael Bograd: Perfect. So I think, you know, your intro honestly teed up exactly what we're going to be talking about. It's going to be all things data centers, right? All things hyperscale data centers. And it sounds like you've been doing, focusing on data centers for eight years, if I heard you correctly. How has that landscape changed?
Bahadir Barbarosoglu: You know, like about seven, eight years ago, when I first did my data center, it was hyperscale, 60 megawatt, 80 megawatt. And over the years, the definition of hyperscale has changed. The landscape, you know, it redefined, well, it has been redefining what hyperscale is, not only from, you know, megawatt capacity ways, but also the pace, you know. It was always called past pace, past pace, but it is faster pace, faster pace than yesterday. So yeah, it's been.
MIchael Bograd: What what do you think is driving that request for faster pace? What what why has that in you know, why have they looked to shrink the schedules when I say they, the owners, why have the owners look to shrink the schedules every single year?
Bahadir Barbarosoglu: Because I mean, looking at all those current demands plus upcoming demand forecasts, doesn't matter what company or institution generates those graphs, you see the exponential growth in demand. So it's almost like, well, if you can turn over a data center today, demand is already there. It will start being operational immediately. It's almost like we are in this game of building data centers, turning over data centers to wishfully catch up with the men.
MIchael Bograd: Hmm, interesting. So, you know, we're both in project controls and we both work on on data center programs. You know, I've I've definitely seen a shift in how scheduling and project controls is approached on a data center. I would have said, you know, a coup you know, five, six, eight years ago when you started, it was it was very schedule driven. It still is very schedule driven, but I think the focus has shifted and I
Load full transcript
think it shifted to more of a a resource driven program. You know, what are your thoughts on that?
Bahadir Barbarosoglu: Yeah, I think, you know, it's always going to be schedule centric, obviously, but. I think the existing constraints are more constraining at this point. For example, the resource constraint was always the subject in discussion, right? But it has become more and more reality regardless what market we are in right now. So data centers in areas maybe closer to metropolitan areas, data centers are everywhere, remote locations. So it's not only the existing resources have been consumed by nearby projects in the current data center locations, those resources do not even exist. Like those resources are not there. know, incentivizing resources to bring them to that market, training existing trade in that market, but also, you know, the reality is like your worker runs to bring resources to that particular area. is a lot fewer than what it was years ago. Same thing with not only human resources, material and equipment. Now again, the demand is high. Everybody wants those material and equipment. do not have, are we building as many manufacturing facilities as we are building data centers? Probably not. So the lead time, lead times is inevitable in that case. So the constraints
MIchael Bograd: Absolutely. Yeah.
Bahadir Barbarosoglu: are definitely, becoming more and more factored in the equation.
MIchael Bograd: so you know we see LinkedIn posts and it says, you know, the top data center constructors, holder is typically at the top of that list can you know regularly. How do you know from a planning and scheduling standpoint? No one to give your secrets about how you get your craft to the site. From a planning and scheduling standpoint, how do you you know, manage that trade resource constraint. How do you reflect it in the schedule? You know, do you are you resource loading schedules? Are you you know are you getting that information from the trade partners?
Bahadir Barbarosoglu: Yeah, we do. It really depends on the client. Admittedly, we do not voluntarily resource load our schedules. It's almost always driven by a client request, which as long as it is feasible for us to get the data, we are happy to support that because we also utilize the outcome of that resource loading exercise. You know, it really depends on the project we are building. If it is a project within an existing program, we do have some you know, as built data past experience that we can utilize from parametric estimating perspective. But you know, it's the data center market has evolved so much, but in a way that now they all look alike. You know, the electrical infrastructure, mechanical infrastructure, you know, the shape, the configuration of the buildings, it's almost like even if you do not resource load, can both. the resource needs. But again, it's never sufficient to go for a constructional baseline schedule, right? We always go and ask our trade partners, well, we are building the subsystem. How many crews make sense to run in parallel, especially for repetitive scopes? And number of crews. doesn't mean anything unless we consider the crew size. So we ask them, okay, so you said three crews, how many people are in each crew? So how long it's gonna take for them to go from step A to step B? So we ran all those analysis. And if we can resource both the schedule, we also look at the head counts. So what we are forecasting, are there any big peaks or downs on, is it smooth or is it volatile? There's some volatility. We do carry those conversations to just challenge ourselves in terms of, okay, is our schedule feasible from resource available?
MIchael Bograd: Fantastic. I I I personally am a big fan of resource loading. I know it's challenging to collect that information, get it into a baseline in time, but I I agree. I think there is a a value add to having that. I think maintaining it is is a challenge though. You know, how do you collect actuals accurately? How do you ensure that the actuals you're collecting, you know, if someone was supposed to work in in data hall one, colo one, cell two. Did that person, did that electrician actually go to that space and perform work in that space? So that's that's always been, I think, maybe you have a different thought, but I think that's always been the challenge with resource loading and maintaining resource loading, is once you assign it to an activity, how do you get those ac accurate actually?
Bahadir Barbarosoglu: Yeah, it's always been a challenge. personally have never been fortunate enough to actualize at the activity level. But you know, there's always, so if we are, you know, managing your project, with a schedule that's at the very, you know, very low levels, very detailed, it's similar to, you know, cost and budget. We can do our estimates at the lowest level. But when it comes to actualization, the data comes at the higher level.
MIchael Bograd: Yeah.
Bahadir Barbarosoglu: So we test this actually, how we can record the actuals in the field. We have the budgeted and earned in the schedule. So what kind of of accounts we can use in the schedule, perhaps activity codes, that can be rolled up to a level that is going to match to the level the actuals are being obtained from site. So it is it's something we are testing right now, but you know even even if we cannot Record the actuals in the field at the level we like the earned value You know generated from schedule based on the budget of labor hours or budgeted commodities It still gives all that we say so if the schedule says well, we have just earned hundred thousand labor hours It may not be a good maybe we earned seventy thousand maybe we you know 200,000 but when we look at what we have earned under thousand labor hours and what the schedule is shopping for us that we are going to be earning 200,000 next month the question is well forget about the accuracy are we really going to progress twice
as much we did last month
MIchael Bograd: Yeah.
Bahadir Barbarosoglu: so you know as long as the accuracy inaccuracy is consistent we can
MIchael Bograd: Sure.
Bahadir Barbarosoglu: still we can still you know scale you know so to kind of poke holes in our schedules.
MIchael Bograd: I gotcha. I love that. Switching gears a little bit, not too much. you have you know you're in a really interesting and unique position, right? You get to see a full data center portfolio, which is pretty awesome. Critical paths of those data centers. You know, the notes I I have from Pedro, he does a really nice job of prompting me, giving me some nice talking points. The notes he's he's letting us know is, you know, power. Power is always going to be an issue on our project, getting power to the job site, getting getting our our data centers teed up for you know all of the testing that's required. Looking at all of your critical paths, you know, for your project or thinking about them. Does it actually run through power?
Bahadir Barbarosoglu: It depends. It depends how late the power is. So if the power is not too late... it quote unquote doesn't run through power. The question is, how much power do we need to commission and partially get this operational? And then when we can actually cut in that delayed permanent. So it's not going to really go through the, there's going to be scopes that's going to be driven by the permanent power, but how we build the schedule, how we commission the building is going to be determined by how much time power we can provide and when, how, and when we can cut in the permanent power. But it's like we're talking about months of difference. if the permanent power is not coming until 2035. it's unlikely for the project to be even executed. So I think, know, again, years ago, 10 power was a question, you know, it was, it was a consideration. Now, if 10 power is, how much, how much 10 power do you?
MIchael Bograd: Tempower's the norm these days. Tempower
Bahadir Barbarosoglu: Yeah, it is.
MIchael Bograd: and you know, rolling and low banking in there
Bahadir Barbarosoglu: And time quick.
MIchael Bograd: as well, all of the above to commission the systems as as quickly
Bahadir Barbarosoglu: I'm the present.
MIchael Bograd: as possible. any downside to that? Any downside to using temp power or have all those kinks been worked out?
Bahadir Barbarosoglu: I mean, there can be several downsides of it. What is our source of temp power?
MIchael Bograd: Yeah.
Bahadir Barbarosoglu: At what gear we are cutting that temp power in. We cannot temp power every single busway. So we will need to pick some sort of equipment that's upstream that makes sense to cut in the temp power. So are we using temp generators? How many generators can we run? What is the local jurisdiction says from environmental perspective? What kind of area those temp generators are going to occupy? What does it mean for our site logistics? It's quite unlike for us to build a single
building on a campus. Imagine there are multiple projects on the same campus. you really need to plan your site logistics in a way that it doesn't cause inefficiencies for the next building. So there are really four or one campus can go with a bigger investment in Tempower, making the decision of, I'm going to build this Tempower infrastructure and I'm going to be using the stamp power for multiple buildings.
MIchael Bograd: Absolutely. and then that also plays into, you know, like early access dates. I feel like, you know, the owners have also this is it's new-ish for me. It might not be new to you, but when I, you know, I I was probably eight years ago as well when I started working really working heavily in data centers. and this early access by the owner was was not really the norm. And I feel like early access is the norm. You know, there's a a very clear transition of the owner's coming in, they're moving in there, they're gonna do their own low voltage terminations, they're gonna start moving in their racks, the space needs to be, you know, positive pressure, conditioned, maintained at that temperature. How does all that play in as well? Does that impact your your power?
Bahadir Barbarosoglu: Well, that definitely impacted how we plan and schedule big time.
MIchael Bograd: Mm-hmm.
Bahadir Barbarosoglu: But from all operational perspective, how we're building, again, years ago, we would turn over a building, a data center, with fewer milestones. And after the turnover date, the building would not necessarily become operational for months, because there's all those scopes that are carried. out after the GC turns over the place, the clients or, you know, other units come and populate tracks and do all those other things. think, you know, owners now they actually fast track certain things. So they introduce more milestones like, you know, asking for certain places to be turned over earlier. So it really, you know, if if you give me a scope and say you have 10, 10 months to finish the scope and I. do my plan accordingly and you come and say, wait a second, actually, I would like you to turn over half of this in first five months. I can make that happen, but the remaining of five months is probably going to take longer than five months now because I turned over half of the space. My access is going to be monitored differently. I would need to keep different housekeeping approaches there. My logistics are going to be constrained. it definitely, I think from overall data center operations perspective, it accelerated a lot. From general contracting execution perspective, it did introduce some more considerations to factor.
MIchael Bograd: I think a lot, yeah. If you're you know, once you that space has hit early access, anyone that's going in and out is is bootied in, you have to clean, there's the the trip the ma dust maintenance, it definitely introduces another level of
Bahadir Barbarosoglu: And also, you know, if you don't turn over that space,
maybe you can, you know, have 50 people working in that space. When it is turned over, even if the spatial, you know, constraints allow, you'll probably not get approval for 50 people. So, yeah, there are a lot of things that factor in that equation as well. But, you know, it's like, and similarly, in the backend of the project turnover, when you have fever milestones, you can get a bit higher level schedules. Now you have your schedules back end is more segregated. You don't have much flexibility. You got to go to the lower level.
MIchael Bograd: I I can picture a handful of scheduled specs. It's I I'm impressed. You know, I think data center owners, you know, what they want in their schedule specs, I think they've all been through the same trials and tribulations that everyone has been. And they've all had this, you know, different bare or varying experiences with general contractors and what they've delivered. And what I'm seeing a shift in is that the the scheduling spec is becoming pretty onerous. You know, it's a prescribed WBS. Prescribed list of milestones. which tells me though, you know, they take scheduling seriously, which I appreciate, you know, because because we you and I are planners. We like doing this. This is important to us. so if they're taking it serious, you know, we have to make sure as general contractors that we're supporting, or you as a general contractor, you're also supporting. So I think, you know, maybe the the you know, when we we look to put bodies staff these projects, and you're putting people on these projects as well as planners. I personally believe it's more than one planner. I I don't know what your thoughts are. I think it's a team that's required to do these jobs.
Bahadir Barbarosoglu: and I cannot think of a planner slash scheduler being able to deliver the product he or she is supposed to while complying with all those requests by himself or herself. It's simply impossible. it's also, even if that person was a rock star and some fantastic
MIchael Bograd: And we do. Yeah.
Bahadir Barbarosoglu: person. Thank you. It is, it defeats the purpose. You know, if the schedule needs to be owned by everyone and everyone should really have a baseline understanding of what we are building, how we are building and why are we being asked to comply with these requirements, right? Like if If the spec is saying, well, try not to exceed activity durations these days, well, there's a logic behind it. Or certain coding, certain elements. And like you said, the specs, the schedule spec, these requirements are being updated frequently or than ever. And well, thankfully, In the past, we would just be handed a spec that we would just have to comply with. Our executives are more aware of those specs, so they ask for subject matter experts' opinion before they commit to those specs.
MIchael Bograd: Mm-hmm.
Bahadir Barbarosoglu: Clients are a lot more sophisticated, so they choose to collaborate with general contractors, get their feedback on proposed changes before they are rolled out. from that perspective, think the industry has has improved quite a bit from collaboration perspective, from baseline understanding perspective of what is planned against scheduling. There's still a lot of room to improve obviously, but it has improved a lot in the past five years or so.
MIchael Bograd: Good, gotcha. You know, sticking with this theme of, you know, you have this really you have access to a lot of projects. you know, procurement is always an issue. O s O CI equipment, CFCI equipment. you know, from a planning and scheduling perspective, you know, where are you s what what pieces of equipment are you seeing at risk the most these days?
Bahadir Barbarosoglu: It is like I would be making it up if I named one piece of equipment right now. Obviously, certain switch gears, generators, depending on the type of busway, it can have a long lead time. Well, still, pretty much those kind of elements have longer lead times now. I mean, it's really procurement. Maybe it was always this important. I came to a late realization. Nowadays, like, it is not smart to just assume we will get the materials and equipment when we would like. So we need to start the chicken and rat situation as early as possible. Like, okay, at day one, as we go through our scope, okay, what is... going to be provided by the client, what we need to procure through our trade partners. Okay, what are the current lead times the market is telling us? Through what funding efforts we can start procuring those materials? What is our turnover date the client is communicating to us? So we start building that puzzle and it's like a hammer. It goes back and forth until it looks together, okay? This is the earliest date, the... site can accept this piece of it. We are not planning to store it. Okay, can we get this on this time? You know what, actually, we're not going to be able to get it on that time. It's going to be two months later than that. Okay, let's take a look at the back end of the schedule. Is there anything we can do? We can accelerate, we can re-sequence. You know, it's... we cannot have a different rigging company for every piece of equipment. So all these material equipment coming to the site that's going to be installed in similar proximity, they need to consider
MIchael Bograd: Mm.
Bahadir Barbarosoglu: this across them as well. know, there's, procurement is really, you know, you know, the chart, like you're to plan as early as possible because that's your biggest chance to influence the rest of the schedule, the rest of the execution. So if we missed that opportunity in the front end, if every PO has been cut and the lead time has already been determined, well, we just need to deal with what we're going to get. We missed our opportunity to influence. So we try to start as early as possible, start with, okay, let's assume everything is available to us off the shelf. How we would build this building. Then
let's take a look at those dates. We need these material and equipment on site. Let's go back to those vendors and owners. Let's see what they can do. Let's plug those dates in. Let's see the impact. Okay. We don't like these impacts. Can we mitigate? Probably not. Let's try to negotiate better. So it goes back and forth. If we are lucky, we can probably finish it in like eight, 10 iterations. If we are not, we will keep having those negotiations throughout the
MIchael Bograd: Let's jump to the very end of the schedule. I think it's everyone's when it comes to a data center, I think this is one of the most challenging parts, commissioning of the space, right? what are your thoughts? How detailed of a commissioning schedule do you like to see?
Bahadir Barbarosoglu: This is
MIchael Bograd: Ha ha ha.
Bahadir Barbarosoglu: a controversial topic. hold on. need to...
MIchael Bograd: Well then.
Bahadir Barbarosoglu: So, my opinion... I would like to keep the commissioning schedule inside my main schedule. having micro schedules, day-to-day plans, fantastic. They are great to have as long as they can be aligned with the overall project schedule. Now, the detail in the overall project schedule is going to be determined by the existing level of detail or vice versa. the existing level of detail will be determined by your commissioning schedule. We cannot have, you know, cherries and watermelons. needs to be pretty much worst case scenario, apple to pears. So, you know, like I cannot really have a commissioning activity per an RP panel while I have a construction activity for electrical branch conduct that's covering 10,000 square feet. It's simply not going to
MIchael Bograd: Mm-hmm.
Bahadir Barbarosoglu: work out. It's not going to give us what we are looking for. It defeats the purpose. We just increase the level of effort to update the schedule. That will take more of our time, plus it will raise more questions. So back to the requirements. The commissioning schedules level of detail. Let's say we have all. the technical experts in the room and in the project organization. So we will not have any problems with, you know, developing, monitoring, updating, skips. I would go and ask, how are we turning over this building? Question number one, if I have one turnover, my commissioning schedule may have less details. If I have 10 turnovers, electrical, mechanical, you know, different, requiring different commissioning levels. My schedule is going to be more detailed. And then what is commission? What is the definition of commission? Because it depends from one piece of equipment to another. Let's say pre-energization checks. It depends from one piece of gear to another. But also it depends from one client to another. So it is really, again, to the back end of the schedule. How is our client wanting to turn over this project? And what needs to be turned over at what capacity? That will determine the lowest detail we can have it.
MIchael Bograd: What I'm starting to see as well is, and maybe, you know, this is I think it's historically been the case too, is the owner, you know, the the general contractor will be involved up to level three or that pre-functional checklist. And then they'll support the level four and level five commissioning, right? And the owner kind of takes it over with their own commissioning agent. So what I'm starting to see on that on the level three is they want it broken out by asset. They want to know it, you know. that granular level because they're gonna have their team getting ready to mobilize and they want to know, hey, if this asset is ready, I need to notify my team, which then to your point is going to say, hey, now in the construction schedule, we need to break out the assets so we can tr map it all so you can trace that path from that commissioning all the way back. Which then kind of gets you into you know that 15,000 activity schedule, which which to your which you were alluding to becomes the challenge stuff.
Bahadir Barbarosoglu: 100%. I mean, those kinds of resources, you said, know, installation checks, pre-functional checks, functional checks and systems tests, integrated systems tests, know, all those things, requirements. It's like an evolution of a race chart. Okay. It's on the trade, the GC monitors. It's on, you know, trade and the owners, subject matter experts and the owner. But if something happens, need to the trade to be there to fix certain things, you know, like It's like starting with fewer parties and then all parties support kind of commissioning approach. But there had been a project in the past where, you know, one of the commissioning agents of the client, you know, he kept asking us, okay, when is this piece of gear going to be available for us to start our commissioning? And, you know, we missed the day for like two days. The day we have been communicating to him, we missed it for two days. It cost us like close to a month because that That particular subject matter, it was a shared resource by a program. So he was already
MIchael Bograd: Mm.
Bahadir Barbarosoglu: booked for the rest of the month. We were supposed to get a building commission. was in a different country pretty much, working on a different project, commissioning a different building. So those kind of constraints, they're still there.
MIchael Bograd: I think those need to be communicated early and often. Those are the r reality of the constraints. Because my head immediately goes to who owns that delay. We don't need to get into that, but you know, hey, you guys missed by two days and now calls a month delay, who's responsible for that, right? That's a that's a big
Bahadir Barbarosoglu: True, also Michael, think about this. So acceleration is not an ask that occurs at certain intervals at this point. Acceleration request is a constant ask.
MIchael Bograd: we submitted a baseline today and the request came in today to accelerate.
Bahadir Barbarosoglu: and you submit
something tomorrow and the day another acceleration
MIchael Bograd: Yeah.
Bahadir Barbarosoglu: will come. And it makes sense. Like I said, the demand is so high, it all makes sense to accelerate as much as safely and feasibly as you can. But when you know this, doesn't matter what they are communicating. What happens if I can accelerate? What happens if that resource is shared by 10 different projects and eight of them can accelerate. Their order has already changed. So in that case, obviously for anything, being able to accurately communicate the data is very important. But having a bench of more resources makes the most sense. think it's the most risky. It's easier said than done, obviously, especially when we talk about resource constraints. But if I had two options, I'm going to spend a lot of efforts to make our days more accurate. or if I am going to spend more efforts building a bunch of subject matter export, I assure you the second option requires less effort and it's going to bring more success.
MIchael Bograd: you know, any you know, to kind of put a bow on this, any any kind of closing words or thoughts for anyone either, you know, in project controls interested in working on a data center project or currently working in the data center project? What's your kind of advice or recommendations?
Bahadir Barbarosoglu: Well, it is such a question. It's a big, sparking question. I can probably have like 15 recommendations and then be
MIchael Bograd: Yeah.
Bahadir Barbarosoglu: regretful about five of them. But let me try. So maybe not specifically the data centers. But when it comes to planning and scheduling, there are thousands of people in the industry, in the construction industry. These people, our people in the industry, should really pay more attention to planning and security. All of them have the experience, all of them have the eye. All of them already have the information that they digest even unintentionally. Everything happens in a sequential manner. Everything. There is one thing on the earth. that is not man-made and it is time. We do not achieve things on time. We achieve things despite the presence of time. So, you know, paying more attention to how things are planned, scheduled, should schedule, should plan in a way that it is future-proof because changes are inevitable is, think, the best investment a person in the construction industry can make for himself. You know, data centers, data centers are great. I love data centers because you know, they're mission critical facilities. I, they're like human beings to me, less, less mobile because you know, they need to be safe, secure, know, cooled and powered 24 seven. And the good thing is how we built buildings. Yes, there are very many tools, more efficient tools, but how the Panama Canal was. constructed versus Empire State Building constructed versus how we are building buildings, there hasn't been much change. Good and bad. So invest in planning and scheduling, you our people should invest in planning and scheduling. That would be one-to-one investment to themselves.
MIchael Bograd: Appreciate that.