Success Factors in Data Center Construction Planning

[00:00:00] Gino Napuri: Thank you for coming today to the NetPoint conference and to the presentation of success factors in data center construction planning. My name is Gino Napuri and I'm here with Ebra Palmer. I will be talking about data center and the early planning needed for successful construction of a data center. The agenda for today is going to be very simple. We're going to have an introduction of the speakers. Then we're going to talk about data centers in general for those of you who don't have much experience with it. Then we'll talk about the different simple data center layout to give you an idea of how the physical building looks like. After that we'll talk about the problems in data center overall and the solution NetPoint. And then we'll do a quick short version of a live version of NetPoint and how it could be used to manage data centers early planning. As I mentioned before, my name is Gino Napuri. I have over 20 years of product control experience focusing on mostly scheduling, air value management, and claim analysis. I have worked in various projects including the Bay Area Clipper Car, automatic fire collection system in the for the Metropolitan Transportation Commission. I also work for the design phase of the California high-speed rail for various sections including San Francisco to San Jose. I work in environmental projects at PG&E. Also some of the commissioning projects for GSK. Also other environmental projects at the Canadian nuclear laboratories. Most recently I work at New York Liberty International Airport New Terminal 1 as part of the project controls team. And last but not least, I work at Google data centers as a coordinator for the East region. I have a advanced use knowledge of the YouTube of Primavera. I'm a certified project manager by PMI and I'm also an air value professional by AAC. And with that, I transfer you to Debra. Hi Debra. [00:02:07] Debra Palmer: Hi Gino. Thank you. And good morning everyone. Welcome to our conference. My name is Debra Palmer and I'm the managing director for PMA's Mid-Atlantic office which is located in New York City. I have more than 30 years experience and I started my career as a general contractor on heavy industrial projects which to this day remains my favorite part of construction. As relates to the tech sector, I was the GC involved in the construction of a 1.8 billion dollar semiconductor processing facility in the state of Washington that had to be constructed and commissioned in 12 months, if you can imagine that. And our client was a Taiwanese consortium. I mention that project because the complexities and timeframe of building a semiconductor processing facility are very similar to the construction of data centers which is what we're going to be discussing today. So just what is a data center? Simply stated, a data center is a physical facility that enterprises, companies use to house their business critical applications and information. Some data centers are just a room in an office. Others are huge, which we'll talk about. Data centers are often referred to as a singular thing. They are not. In actuality, what they are is they are composed of a number of technical elements that can be broken down into three categories. The first category is compute. The memory and processing power to run the applications generally run by high-end servers. The second category is storage. Important enterprise and business data is generally housed in a data center. And the third category is networking. Interconnections between data center components and the outside world. All data centers are essentially buildings that provide space, power, and cooling for network infrastructure. They centralize a business's IT operations and equipment as well as store, share, and manage data. And this flow diagram that you're looking at is a good example of how it works. Next slide, please. So it should go without saying there is a growing demand for cloud applications and the technology. And this has become a key enabler for business, social connectivity, and entertainment. We all are dependent on the cloud now. Accordingly, technology firms have been forced to fast track data center construction, expansion plans, and even upgrades to meet the needs of the end users, which are all of us, which are all of us. We all use this now. And a good example of this fact is that Microsoft, who currently operates more than 200 data centers around the world, is now on track to build between 50 and 100 data centers each year for the foreseeable future. It's a perfect example of the growth. Next slide, please. So the most common data centers, and there are so many around the world. While most are small, the average data center occupies approximately 100,000 square feet of space. On the other end of the scale, there are behemoth data centers that consume as much power as a medium-sized town. It's almost impossible to imagine. The United States is the home to the highest number of data centers. We have just over 2,600, or 35% of the global total. There is a considerable gap when you go to the second place country, and that's the UK with just 451. Germany comes in as a close third with 442. Next slide, please. And then you've got the big babies, the hyperscale data centers. Worldwide, there are 500, 44%, as you can see, are in the United States, 8% in China. Hyperscale data centers are massive business-critical facilities, and the largest hyperscale data center in the world is 6.3 million square feet of space. Remember, I said the average was about 10,000. 6.3 million square feet of space, and it's located in China. Next slide, please. So what are the critical success factors in data center construction? Given this demand for the construction of more and more data centers, and it is a demand, it's very interesting to note that the data center construction pipeline, this is an actual organization that monitors this. The data center construction pipeline reached near record levels by the end of 2020, totaling 611.3 megawatts in the United States alone. And that was during a pandemic, if you can even imagine. Further, the data center construction market is expected to grow at a compound annual growth rate CAGR of 8.34%. Again, 8.34% during the period 2021 to 2026. Given that you figure, and it's true, the demand and the need is great, and time is money, as that hackneyed phrase goes, it's so true here. And so to ensure the successful construction of the data center, the owner and the contractor must, must, focus on the factors that will make the project a success. And here is the list: community awareness, procurement transparency, use the A-team, schedule sequencing, defining the critical path, stakeholder buy-in, and of utmost importance, along with these, next slide please, is safety. Safety in the data center construction market is paramount. Safety is always critical on a job site, but particularly in a data center. So, the safety rules, you assess the risk, control hazardous energy, assure electrical safety, working at heights, hot work permits, lifting and handling, equipment that's being placed. There are, if you even look for it, Google it, I did so, some amazing safety procedures that have been developed specifically for the construction of a data center. In addition to the safety and the previous critical success factors, and speaking as a long-time project manager, it's critical to avoid four common project management problems when constructing a data center. And we've all shared these, those of you who are out, sitting out there that are project managers or owners, but again, four common project management problems that need to be avoided when constructing a data center. And I personally had a lot of issues with this when I was doing the semiconductor processing facility, and that there was no labor readily available, and we had to bring in people from other states, and the productivity went down tremendously, and that causes an impact to the schedule. And number four, lack of change management readiness. So what Gino and I are talking about today is NetPoint and the success factors in data center construction planning. So point number two, keeping the project on schedule, is where NetPoint becomes an invaluable tool. And with that, I will now hand the presentation back to Gino, who will discuss the use of NetPoint in data center construction. He's brilliant at this, so pay close attention. [00:11:21] Gino Napuri: Thank you, Debra. I really appreciate it. So let's start from the concept of how the data center looks like. Let's talk about the layouts. This is a simple layout of a data center with its main components. We're talking about the main server, the main building, which contains the data center hall and the CNR, or core network room. Of course, to power all this, you need a substation. We provide the energy, the electricity, but you always need redundancy and backups. And for that, you have generators that are sitting on the side of the data center, usually. And these generators are energized with diesel fuel from diesel fuel tanks nearby. Also, for the water to get cool, you need a mechanical cooling building and cooling towers. And last but not least is a chemical treatment building where all the chemicals get treated before disposal. For this data center, this is a simple layout of the data center. There are many more components, but these are the main components in most data centers. And this is how they look like. Inside mechanical cooling building, you see the chillers where the water gets processed and cooled. Also, you have in the central network room is where you see the switches and all the information comes into the central network room and get distributed to the server halls that look like this. The server hall has different clusters where the information, where the servers, where the information gets processed. Outside, you have cooling towers that cool the air. In this picture, they're up here, big giant fans, and also water tanks to help those cooling towers. Last but not least is the substation. In this particular picture, this substation is partially solar power that require energy for the data center to function properly. So now we're going to talk about the problems in data center planning. At the early stages of when an owner or a client wants to develop a data center, what are the issues and consequences? But before that, we also want to talk about considerations about selecting the data center location. Not just the location is key to build a data center, but also the weather impacts the decision to build a data center. The colder is outside, the less energy you need to cool down the data center inside. Other issues include the possibility for local governance or vendors to provide enough energy, internet speed, and tax incentives to build a data center for the companies. The last thing I want to mention here is that data centers can be built over basically in any kind of field. It could be greenfield, brownfield, or even existing buildings and warehouses. They can be used as shell for the data center. So let's talk about the issues at hand. There is information that needs to be passed, and the communication is very critical, and sometimes people lack good communication skills, or the communication doesn't go upstream or downstream unnecessarily as quick as possible for the whole team to be in the same in the same lane. Reporting issues. Sometimes the reporting itself, the accurate information within the reports is hard to find, and they're not aligned with the deliverables. Timing. There is issues that because the stakeholders are in different sections of the organizational chart are receiving information that early or late information or changes occur rapidly, and the whole tree doesn't understand of the changes as they occur. And accountability. Lack of transparency in communications and accountability by members in all sections of the organization causes less reliability and trust and the data being shared. So where's the solution for all this? We have NetPoint. Because why? NetPoint is a single source of truth for the information being shared in regards to scheduling for the early planning data center. It's a team effort. It requires all the stakeholders to be involved from the beginning through the end of the maintenance of the schedule, and it helps everybody be aligned and informed at the same time. This is the reporting. Whatever changes are made in NetPoint are actually seen automatically, which help you have dynamic results. Whenever you make changes and the changes affect the finish days for your project, then it's when you become more aware of what's going on and your team is able to perform once in sync towards the completion of the project or towards the baselining of the project. How do you get a schedule developed? Usually you start a project with the location planning and coordination team selecting a site depending on the factors that we talked earlier. Once the site is selected, then you go and distribute the information to the site team to develop a NetPoint schedule, a summary schedule, we'll say here. If the site is a brand new site, then you develop the NetPoint schedule in a new canvas in NetPoint. But if the site is an existing, is the projects in an existing site, then you add the project to the already existing canvas in your schedule. All this is done early in the project before any funding for construction is given. Usually the owners divide their funding for the data centers into various stages, from early planning to actual construction phases. So if the project is fully funded for it to be built, then you go ahead and develop and maintain a detailed schedule. But before that, when the resources are limited and the information doesn't have to be that detailed and granulated, then you can go ahead and manage your schedule using NetPoint. Here's an example of a template for a data center of two phases in NetPoint. As you can see, it is broken down in three main subdivisions, but these divisions have also other sections that are very crucial to the maintenance and management of this summary schedule. First, we have the early planning, where the location energy planning team selects the location for the actual data center to be built, the land gets acquired, and you receive initial funding to do some early design and permits. The second stage is where the more permits are required, the design is completed, and the foundation can start being built for the data center location. Another part is the procurement. Basically, this is the four main components. So the main components required for a data center to function and things that need to be delivered on time for the construction schedule not to be delayed. And finally, we have the actual construction stages of the project. In this particular project, you have a substation being built at the same time as you have a construction phase one and construction phase two being built. And all these are connected to the construction complete phases that we will discuss later. But this is basically a nutshell how your NetPoint schedule is broken down into these sections to manage. As you noticed in the previous slide, there was a critical path that started only as the construction funding approval. In this example, we noticed the critical path being once construction funding occurs. This is a more realistic approach to the plan as funding tends to be key determinant to the construction stations of the project. The greatest advantage to use NetPoint is that we use plan dates to set the activities. This means that we are not considering constraining to use the early dates like other scheduling tools, giving us a more realistic plan for the project. This particular template reflects the "hurry up and wait" approach that some data centers, construction management teams have to deal with every day. So we're watching in the previous slide one project. What happens if you have multiple projects? This is how NetPoint is created, can create two projects in one canvas to reflect the progress of your schedule and interlinking between projects. In this particular project, I have to finish the final design of project one, be the predecessor to the preliminary design of project two because it is believed that the design team will be the same thing doing both projects. Here we have project one highlighted for better identification in the same canvas and here we have project two. Now this is a level two schedule, as you can see, with the details necessary to manage a project at a high level. So what happened, how does this look at a level one? Here's an example of three projects at level one. As we can see, the project number one is located right here in the level one summary with the three main components: front-air planning, procurement and main construction. Project two has the same components next to it and project T has the same components at the bottom. In this particular example, I also want to show you how this looks on the actual site. The project one is a project that's already almost completed, as you can see on this section of this map. Project two is halfway completed with being finalized but not completed yet. And project three is here on the early stations with the still direction is starting and that and that can be managed all at one level in NetPoint. So with that we go ahead and going to go live with the tool to play with it a little bit and see how NetPoint works in real life. So here we have a level one schedule, a level two schedule, a summary and let's say and we have the main components that we discussed earlier. We have the early planning, the design, the procurement and the construction. Something that I wanted to mention as well was the stations right here which is the the three main milestones for any data for most data centers when they're being built. The construction complete is where the general contractor is done with activity with other construction activities in the project. Equipment installation completes without the hardware, all the servers and all the routers and all the switches are installed on the data center. And phase one ready is when the machine learning is complete and the and the much and the machines are ready to and the server is ready to be used. As we can see here if we have an impact in any of the activities here the milestone will shift. So we have final design being delayed for example instead of 142 days they say it's going to take 180 200 days. You can see how the schedule changes and all the stakeholders are aware of the new impacted of the new impact created in that point instantly. We can also say something that also delays the project in a regular basis is the manufacturing. When the generators or the manufacturing of the generators get delayed because some of the materials are not available then you also have to make those impacts reflect on your summary schedule and you can do that by just sliding the bar towards the duration that you draw the new delivery date and this the whole team can realize how the how the schedule impacts the finished milestones. This is a view of one summary schedule and one one one net point schedule for one project in one site. How does this look on multiple projects in the same site? This is an example of multiple projects on the same as you can see they're connected they're interconnected in key in key activities. We have as mentioned before the final design be the predecessor for the preliminary design between project one and project two and the same team between project two and project three. But we also hear from the construction being limited between the project one and project two phase one and break one break two phase one and break and also between project two phase two phase one. This could be reasoning again as mentioned for the resources. You only have certain availability of resources when it comes to design but most importantly construction. When you're in the construction stages of the project there is a need for resources and if you're working building the same data center in the same location you will most probably reach a maximum of resources when it comes to electrical engineers and that's why you need that relationship. And again, if I delay the design on one project, all the projects gets impacted and the new finish dates for other projects get updated automatically and this information can be shared with the team when required. Same thing with construction. If a phase one construction gets delayed, the schedule shows those changes as we can see here automatically and the team can be aware of how the decisions affect the overall project and the flow of the construction stations at a high level.