Designed with STAR-CCM+: The world’s best bottles. · From the blog: Flo Cycling wheels 4-5 From the blog: Question and learn with data focus 6-7 Bottero: Stronger, lighter glass

siemens.com/mdx

Designed with STAR-CCM+: The world’s best bottles.

STAR-CCM+: Discover better designs, faster.Improved Product Performance Through Multidisciplinary Design Exploration.

Don’t just simulate, innovate! Use multidisciplinary design exploration with STAR-CCM+ and HEEDS to improve the real world performance of your product and account for all of the physics that it is likely to experience during its operational life.

siemens.com/mdx

1

Introduction | Dynamics 41

Jan LeuridanSenior Vice PresidentSimulation and Test Solutions, Siemens PLM Software

Welcome to Dynamics 41, which is the first edition under the Siemens brand.

Siemens provides a comprehensive portfolio of seamlessly integrated software, hardware and technology-based services in order to support companies worldwide in enhancing the flexibility and efficiency of their engineering and manufacturing processes and reducing the time to market of their products.

Sometimes, when we talk specifically about all the ways that simulation technology can help improve the performance of products in high technology industries such as automotive, aerospace, electronics and life-sciences, it is easy to forget that engineering simulation now touches every industry. Indeed, the cover story illustrates how some of the very oldest industries are now turning towards predictive engineering analytics in order to offer new insight into the development of their products and processes.

As far as multidisciplinary simulation challenges go, predicting how a molten lump of glass is formed into a structurally rigid (although still technically liquid) glass bottle is perhaps one of the most complex. Manufacturing a glass container involves all modes of heat transfer and structural and fluid mechanics of a material whose viscosity changes by seven orders of magnitude as it is molded, formed and blown into a bottle or jar. Despite this complexity, the glass manufacturing industry has, until very recently, remained untouched by the world of simulation. Instead, the manufacturing processes behind a glass bottle are based mainly around decades (or even centuries) of accumulated engineering knowledge stretching to before the industrial revolution.

Introduction

It’s not that this accumulated engineering expertise is wrong, it usually is not, but it ’s almost always based on incomplete knowledge of the processes involved. Engineering simulation lets us see inside the black box, both in a literal sense by visualising the physical processes that occur inside previously unseen spaces (such as an engine cylinder or inside the mold of a bottle making machine), but also by granting us a deeper understanding of all the physics that influence the product or process in question. Engineering simulation builds on traditional engineering expertise and allows engineers to take the next step in designing the products and processes of the future.

In this case, the payoff is bottles that are not only stronger and lighter than ever before, but also cost less to manufacture and consume fewer raw materials. Indeed, some of the bottles designed using predictive engineering analytics were so strong, that they exceeded the capabilities of the machines used to test them.

Enjoy your read!

2

Dynamics 41 | Contents

Bottero: Stronger, lighter glass bottles designed with STAR-CCM+ Full story on page 8

Editorial Dynamics welcomes editorial from all users of CD-adapco software or services. To submit an article: Email: [email protected] Telephone: +44 (0)20 7471 6200

Editor Anna-Maria [email protected] Associate Editors Christopher [email protected] [email protected] [email protected] [email protected] [email protected] Design and Art Direction Ian [email protected]

PhotographyFranco [email protected] Press Contact Americas: Todd [email protected] Europe: Julia [email protected] Advertising Sales Geri [email protected] Events US: Lenny O’[email protected] Europe: Sandra [email protected]

Subscriptions and digital editions: Dynamics is published twice a year and distributed internationally. All recent editions of Dynamics, Special Reports, and Digital Reports are available online: http://mdx2.plm.automation.siemens.com/magazine

3

Contents | Dynamics 41

Dynamics 41 | Contents

Introduction 1

From the blog: Flo Cycling wheels 4-5 From the blog: Question and learn with data focus 6-7 Bottero: Stronger, lighter glass bottles 8-12

Simulating a conceptual FLNG unit in waves with STAR-CCM+ 14-19

Trends in CFD applications for the maritime industry 20-23

Know the ropes - The America’s Cup wing sails 24-29

Contributing to the efficient use of electricity and heat in power generation 30-35

Driving battery innovation with CAE 36-40

Ford Otosan improves diesel engine design with STAR-CCM+ 42-47

Improving cooling effectiveness of gas turbines through 48-55 design exploration

CFD simulation for a fully featured offshore platform model 56-59

Savvy separators 60-67

Degassing Africa’s Lake Kivu for public safety and power 68-72 generation Multi-scale modeling of furnaces and reformers 74-81

Prime the pump - Introducing simulation-led design 82-88 exploration to centrifugal blood pump development

The value of CFD in respiratory medicine 90-94

Modeling additive mixing in weir flows 96-99

When the wind blows... 100-103

4

From the blog | FLO Cycling wheels

Figure 1: The Flo Cycling wheels It starts with an idea. Morphs into an obsession. Then comes the sweat. Practice. Tears. You hear the doubters. It’s an obstacle course. Persevere. Focus. The pressure is on. Determination. Your body and mind are beaten. It’s an arduous journey. Then you see it. The glorious finish line. The reward. The victory.

I’m talking about triathlons here. But I could very well be talking about something else with similar challenges – Startups. Triathlons are intensive and grueling and definitely not for the weak-hearted. Just like shepherding a startup company to success. This is probably why I’m always doffing my hat in respect for Chris and Jon, brothers, triathletes and the founders of FLO Cycling.

Three years ago, we ran a story in Dynamics about FLO Cycling, a startup company poised to crash the bike wheels party with their highly aerodynamic racing wheels at low prices, inspired by the brothers’ search for a leading-edge design at affordable cost. This week, we were thrilled to hear back from the “Bicycle Brothers” and guess what? FLO Cycling has arrived to the party – in some style. In early 2009, the market for racing wheels offered two options: highly aerodynamic,

Designed with STAR-CCM+: FLO Cycling wheels

wind tunnel tested wheels at a high cost or affordable lower cost wheels with significant performance reductions. When Chris opened up his new set of highly aerodynamic (and expensive) wheels in 2009, the brothers realized that quality and affordability don’t always need to be mutually exclusive and thus, FLO Cycling was born. Their initial offering of aluminum/carbon wheels, designed with STAR-CCM+® Software was a resounding success. The racing market is moving towards carbon clincher wheels, both for their versatility and performance characteristics. To address this, FLO Cycling is offering a completely redesigned, all new FLO Wheel product line.

To bring these wheels to life, they adopted a five step Data Driven design process that took 15 months from concept to final design.

1. Data collection To design better wheels, it is crucial to know the riding conditions that the wheel will encounter. So what did they do? They built a custom data logger connected to two custom sensors, one for yaw angle and one for relative velocity, and mounted it to the front of their bikes while competing in four

5

FLO Cycling wheels | From the blog

Designed with STAR-CCM+: FLO Cycling wheels

From the blog Prashanth S. Shankara - Siemens PLM Software

Ironman races. Insane? Yes. Inspirational? Absolutely.

2. Data analysis With over 100,000 measurements from their data collection, they sifted through the numbers to find useful patterns for the design. The data analysis process uncovered relationships between relative velocity, yaw angle, and time for multiple riding scenarios (coastal, inland, drafting to name a few). Armed with this information, they identified the percentage of time spent in various yaw angle ranges. Contrary to their belief, it was realized that the majority of time (over 80 percent) is spent in the 0 to 10 degree yaw angle range and this narrowed down the design space for the new wheels.

3. 3D modeling FLO Cycling partnered with the Engineering Services team at Siemens PLM Software for a simulation-led design process. The base for the CFD optimization study was the Continental GP 4000 S II tire, aerodynamically the fastest tire available. Using a combination of mold and Autodesk Inventor, an accurate 3D model of the tire was created capturing the details of the tire/rim interface accurately.

CFD optimization A custom optimization algorithm was created to identify the fastest rim shapes from the STAR-CCM+ simulations. The Engineering Services team at Siemens used STAR-CCM+ to iterate through 500 different prototypes in two months, with each design evaluated at four different yaw angles (2.5, 7.5, 12.5 and 17.5 degrees). The entire process was automated using Java macros.

4. Wind tunnel testing For confirmation and further evaluation, the finalized designs were tested in the A2 wind tunnel in North Carolina, along with testing of the 2012 wheels. The results from the wind tunnel confirmed the performance of the new bike wheels as predicted from the CFD optimization algorithm. The four new wheels are faster, lighter by a pound, and more compliant on the road.

In addition to this, the brothers share their entire story and the design process in detail in their blog at flocycling.blogspot.com. Are you a cyclist looking for fast, aerodynamic wheels at affordable prices? You can check out the shop at http://www.flocycling.com/store/index.php.

6

From the blog | Question and learn with data focus

Figure 1: Flame temperature

contour plot

Socrates…not a name that comes up in everyday conversation. Yet, as CAE practitioners, let me propose that we apply the Socratic method every time we analyze our simulation results. The Socratic method, or elenchus, “is a form of cooperative argumentative dialogue between individuals, based on asking and answering questions to stimulate critical thinking and to illuminate ideas and underlying presumptions” (from Wikipedia). So, who is having this dialogue? Perhaps we could consider the numerical simulation to be our teacher in this exchange. True to the Socratic method then, the better your questions are, the more you learn. What does this have to do with Data Focus? In short, our latest enhancement, Data Focus, gives you an unprecedented way to ask your teacher (the simulation) really good questions and get great answers.

When reviewing simulation results, we commonly reduce informational detail, narrowing our focus, to get a better understanding of what is happening. In the example provided, we see four illustrations, all depicting flame temperature.

In figure 1, the flame on the left is colored by temperature, in full detail. Let’s say we

Question (like Socrates) and learn with data focus

want to focus on where the temperature in the flame is high. We could clip the temperature scalar field at the displayer level, shown second from left (the minimum temperature is set to 1500ºK). In this illustration however, we’ve lost some of the context – the complete flame structure is less clear and the colormap is re-scaled to fit the modified temperature extents. By applying a Data Focus object, shown in the illustration second from right, we highlight the part of the flame above our minimum target temperature of 1500ºK. The original colormap-to-temperature mapping is preserved and values below our minimum have the color hue removed, thereby retaining the visual context of the flame. But, we can take this one step further and ask a deeper question: What part of the flame is above 1500ºK and has a mixture fraction, f, greater than 0.5? The result of this Data Focus is shown in the rightmost frame.

So, what is a Data Focus object and how can you make one? Let’s start by introducing the “Heatmap,” a new style for XY plots, designed to manage heavy plot data when the content would otherwise overwhelm the symbol/line style. Essentially, we use the plot axes to define

7

Question and learn with data focus | From the blog

Question (like Socrates) and learn with data focus

the rows and columns of a spreadsheet. Each cell defines a “box” whose upper and lower limits are based on a small range of both the vertical and horizontal axes. We fill up the boxes, matching our input data to the bounding range for each cell. Some boxes will be empty. We can get a sense of how full each box is by using a scalar function to weight the results. If we color each cell in our spreadsheet based on the weighting, we see how our data is distributed within our XY plot. The end result of this numerical manipulation is our “Heatmap,” see figure 2.

The input for our XY plots can be any combination of derived parts, boundaries and/or regions – just drag them straight into the plot window. For our flame simulation, we have the following two heatmaps, one showing temperature versus mixture fraction (weighted by temperature) and the other showing mass fraction O

2 versus mixture fraction

(weighted by mass fraction O2).

We begin by interactively creating a Data Focus object, or selection filter, dragging a rectangle onto our heatmap to focus on temperatures above 1500ºK. Next, we create a compound filter by applying the first Data Focus filter to the mass fraction of O

2 versus mixture fraction heatmap, and

add a second selection filter focusing on the mid-range of the mixture fraction. Since these XY plots are linked to the same Data Focus object, a change to either selection is applied to both. Lastly, we apply this Data Focus filter to our scalar displayer showing the flame temperature at right in figure 3.

To see whether there is a part of the flame where the temperature is high and the mixture fraction is also low or high, you interactively drag the selection in either XY plot – the other XY plot and the scalar displayer get automatically updated. And consider this: For a single Data Focus object, you can create a compound set with multiple filters, no limit. To that

single Data Focus object, you can add any number of multiple compound sets, via interactive selection, no limit. You can create any number of Data Focus objects, no limit. You can then apply any one of these Data Focus objects to any displayer within a scene, or to an XY plot, no restrictions. Simply put, there are no limits to the number, or complexity, of questions you can ask. Talk about the student calling the teacher to task!

I believe we run simulations because we want, and have the strong need, to learn. We want to design better products and we want to be certain of what changes cause what effects. To learn, we need to ask questions and Data Focus delivers an absolutely unprecedented and competitively unique way to do just that. I’ll close here by paraphrasing Plato – “The unexamined .sim file is not worth running.”

Figure 2: Data focus plot of

temperature (above left) and

mass fraction (above right)

versus mixture fraction

Figure 3: Data focus with clipped

temperature and mass fraction

range with flame temperature

contour plot (below)

From the blog Matt Godo - Siemens PLM Software

8

Marine | Simulating an FLNG unit in waves

9

Stronger, lighter glass bottles | Manufacturing

As far as multidisciplinary simulation challenges go, predicting how a molten lump of glass is formed into a structurally rigid (although still technically liquid) glass bottle is perhaps one of the most complex. Manufacturing a glass container involves all modes of heat transfer and structural and fluid mechanics of a material whose viscosity changes by seven orders of magnitude as it is molded, formed and blown into a bottle or jar.

In a deeply traditional industry, one company is revolutionizing bottle manufacturing by deploying multidisciplinary simulation to understand exactly what happens inside the bottle making process and using that information to build better bottle making machines.

Bottero S.p.A is an Italian company that specializes in making machinery for the manufacture of various types of high quality glass products, including a “hollow glass” division that designs and manufactures bottle and container making machinery. Bottero’s aim is to allow their customers to develop innovative new lightweight glass products that are structurally superior to previous designs (and therefore more durable), but can be manufactured using less raw material and

less energy, both to melt the glass and to transport the final container. Ultimately, this creates a quality product at a lower overall cost.

Put simply, Bottero is using multidisciplinary simulation to discover how to make better bottles, faster than ever before. I spoke to Marcello Ostorero, Bottero’s Head of R&D, who pioneered the use of engineering simulation at Bottero, and Simone Ferrari, who performs many of the STAR-CCM+® Software calculations at the heart of the simulation process.

The challenge: Making stronger, lighter, bottlesAlthough we think of glass as a solid, it is in reality a supercooled liquid, whose viscosity is so great that its molecules do not move freely enough to form crystals. Managing the way that glass flows and is cooled to its (near) solid state is critical in ensuring the strength of the final container.

In simple terms, a glass bottle is formed by molding a glob of molten glass (enough to make a single container) into a preliminary bottle shape known as a parison. This parison is then carefully cooled while being blown into the final bottle shape by

Designed with STAR-CCM+:

Stronger, lighter glass bottlesBottero S.p.A is changing the way bottles are being made and they

are using simulation to do it

Stephen Ferguson - Siemens PLM Software

10

Manufacturing | Stronger, lighter glass bottles

a stream of compressed air before the bottle is subjected to a number of downstream processes.

“Our aim is to make lighter bottles, that use less raw materials, less energy to melt and therefore cost less to manufacture. However, since glass is a very sensitive material, we also have to ensure that the bottles are very strong,” says Simone. ”Not breaking is the most important thing that a bottle has to do.”

In the past, the robustness of glass containers was ensured by over-engineering them to some extent, thickening the walls of the containers by adding more glass. However, this resulted in heavier products that were less consumer friendly and more expensive to manufacture. In the past 20 years, thanks to developments in manufacturing technology and to the combined influence of consumer preference and economic necessity, the weight of a typical glass bottle has reduced by over 40 percent without any loss in structural rigidity or increase in fragility. As we shall see later, modern lightweight bottles are often much stronger than their older, heavier counterparts.

“In order to make a structurally strong bottle there are two critical stages: In the

first step a glob of molten glass is molded into a parison, which is a preliminary bottle shape. After this the final bottle is formed in another mold,” says Marcello. “Getting this first shape right is extremely important in ensuring the structural strength of the final bottle; it has to be very precise otherwise the bottle will break during normal usage conditions.”

During the manufacturing process the glass is cooled from over 1,000 oC to ambient temperature, during which time the viscosity of the glass increases by seven orders of magnitude (from 100 poise (P) to 1.0e9 P). If the bottle is cooled too rapidly or unequally, then internal stresses are generated in the walls of the container that reduce its overall durability.

The significant problem in this regard is that it is impossible to understand what actually happens to the molten glass during the molding process, which happens unseen inside the bottle making machine. Historically, the only way to judge the effectiveness of an extremely complex physical process was to look at the quality of the final product, its glass distribution and try to imagine what might have gone wrong inside the mold.

“The strength of the final product depends to a great extent on how the glass is cooled

The history of glass containersWe take glass bottles, and other glass containers, for granted. Sometime today, between now and when you go to bed, you will reach into a cupboard or refrigerator and pull out a glass container, containing food, a condiment or most likely some type of delicious, possibly alcoholic, beverage. Glass is, and always has been, a key component of modern life.

Glass bottles are the signature of a quality product. If you don’t believe me, then try taking a plastic bottle of wine to the next party you attend and observe how other guests react to your gift. Companies such as Coca-Cola, Perrier, Orangina and Heinz have built entire brands around the distinctive shape of their glass bottle. While cheaper and lighter plastic alternatives are available, nothing protects and preserves the contents of a container as well as glass does. The oldest drinkable bottle of wine in the world dates from 1727. A Finnish sommelier recently had the opportunity to sample a bottle of 200-year-old champagne. This is how she described it: “Despite the fact that it was so amazingly old, there was a freshness to the wine. It wasn’t debilitated in any way. Rather, it had a clear acidity which reinforced the sweetness.”

Almost uniquely in a world that is largely blind to the environmental consequences of “disposable packaging,” glass is both reusable and recyclable. Glass is 100 percent recyclable, it can be recycled again and again without any loss of quality or unity. Recycled glass is also an important component of manufactured glass - significantly reducing the energy demand of the process. For example, in 2012 96 percent of the glass bottles sold in Switzerland were recycled.

We have been using glass to make containers for food, drink and medicine for over 7,000 years. Although we have refined the glass making process to the extent that an individual factory can turn out two or three million bottles a day, the basic process remains the same: take a molten lump of sand, soda and limestone and make into a bottle through a multi-step process of molding, forming, blowing and annealing.

11

Stronger, lighter glass bottles | Manufacturing

during the manufacturing process,” says Simone. “Although we can measure the temperature of the mold in the glass plant, without simulation we have little or no insight into the actual temperature of the glass itself. The standard approach in the industry is one of trial and error. In order to perform trials, you need to stop the manufacturing process, sometimes for months at a time. This costs time and money, and doesn’t give any real insight into the problems in the process itself.

A large bottle manufacturing plant can produce more than two million containers a day, or 25 bottles per second. The cost of these “trial and error” investigations or unresolved problems in the manufacturing process is huge. For this reason Bottero decided to deploy engineering simulation as a way of gaining detailed insight into the bottle making process, performing simulations that improve both the process itself and the quality of the glass containers produced.

The solution: Multidisciplinary simulation using STAR-CCM+“To make a good glass container, you need to actually study the physics of the glass,” says Marcello, relating the key insight that is at the heart of Bottero’s simulation philosophy. “Rather than thinking only of the mold, we oriented our view on the glass itself. The big advantage of the simulation is that it allows you to really understand what is actually happening inside the mold. You cannot see that from physical experiments because the mold is closed, it’s made from cast iron, so you can’t see what is happening inside.”

The glass forming process is also extremely sensitive to changes in machine timing, glass composition and environmental

conditions. Simone explains: “As it is nearly impossible to physically visualize what really happens inside the molds during the different phases, numerical simulation is the only tool available to help better understand the details of the physics as they occur during the process.”

“We have very complex physics,” explains Marcello. “If we look just at the machine production, structural and fluid dynamic aspects can be separated. If we look at the product we have to manage, they can’t. They must be treated together. They are very, very coupled. We produce machines to make containers. The container is made by the cooling down of the molten glass, but it has very hard structural requirements. Understanding the actual temperature of the glass is by far the most important factor in ensuring the strength and quality of the final container. Multidisciplinary simulation using a tool like STAR-CCM+ is the only way that we can achieve that.”

Marcello continues: “To improve a glass container using trial and error alone can involve many weeks of lost production. We can achieve the same thing using simulation in less than a day.”

However, solving the engineering problem is not the only challenge that Marcello and his team had to face. The glass making industry is deeply conservative, sometimes relying on experience gathered over decades and passed down over centuries. Although this experience-based knowledge is always valuable as a starting point for developing new products, it is unsuitable for the sort of intelligent design exploration required to facilitate true innovation.

“Many, many, many people from glass plants tell us, ‘Oh, it is impossible to simulate the structural resistance of the

12

Manufacturing | Stronger, lighter glass bottles

glass. Oh, it’s impossible to simulate the forming process of the glass. You are crazy,’” says Marcello. “Maybe at the beginning, we were crazy. But now we are consistently obtaining good results and simulation is allowing us to discover important new markets in which we have no competition for the moment.”

The payoff: Stronger, lighter bottles that exceed customer expectationsHow much better are the products designed through simulation?

“Well, for example, not many people realize that the bottle for a sparkling beverage must resist to at least 13 times atmospheric pressure,” says Simone. “At the same time, we are trying to reduce the weight of the container, save money, and increase the structural performance of the vessel. We can only do that by controlling the glass distribution in the container so that it is as close as possible to the ideal, and thereby avoid defects in the glass.”

“Recently we helped one of our customers to reduce the weight of a bottle that holds carbonated beverages using simulation,” continues Marcello. “When he tested the

bottle that we helped him to produce, he was unable to make the bottle explode. This is an incredible achievement - simulation has pushed the structural performance of the bottle beyond the capability of the testing machines. This would have been impossible without simulation.”

“None of our competitors make extensive use of simulation,” continues Simone with a smile on his face. “So this has been really, really good for our customers. They are starting to pay us to show them how to improve their manufacturing, for example, how to design better molds and ultimately produce better bottles. So we are actually acquiring knowledge from simulation that is superior to that previously gained by experience alone.”

An eye-catching bottle that sets itself apart from the crowd can be a conversation starter at a gathering or just something that is fun to look at. Bottero understands this and realizes the importance of a creative bottle design to the consumer. By using simulation in the manufacturing process, Bottero satisfies consumer demand for a unique product by producing a bottle that is well made, lightweight and leaves a lasting impression.

13

STAR-CCM+: Discover better designs, faster.Improve the thermal design of your electronics systems with rapid, automated design exploration.

siemens.com/mdx

14

Prelude FLNG Hull Float Launch, Geoje, South Korea, 2013 (image courtesy of Shell International Ltd.)

15

Simulating a conceptual FLNG unit in waves | Marine

Max Haase and Yuting Jin - Australian Maritime College | AMC Search Ltd.

Simulating a conceptual FLNG unit in waves with STAR-CCM+

Introduction475 kilometers off the western coast of Australia, Prelude FLNG, the world’s first floating liquefied natural gas platform, is about to revolutionize the way natural gas is produced. As the largest offshore facility ever constructed, Prelude FLNG boasts a length of 488 meters, a width of 74 meters and weighs around 600,000 tons.

Still in its early days, the floating liquefied natural gas (FLNG) technology will allow the freshly extracted natural gas to be processed and stored aboard before being loaded onto LNG tankers, thereby permitting the exploitation of offshore resources that had been too costly or difficult to develop otherwise. In a scientific study undertaken by the Australian Maritime College (AMC) – a specialist institute at the University of Tasmania that focuses on seafaring and maritime

engineering – numerical simulation was used to investigate how various wave scenarios will affect the motions and operations of such a facility. The computations were performed using STAR-CCM+® software.

The Prelude FLNG project, initiated by a consortium in which the energy group Royal Dutch Shell is the majority shareholder, is the first of its kind. In principle, the FLNG processing units are similar to the FPSO facilities (floating, production, storage, and offloading) used for oil extraction, although Prelude FLNG will work on a much bigger scale. The natural gas produced at the field will be cooled to -162°C, at which temperature it turns into a liquid and its volume is reduced by a factor of 600. The liquefied gas can then easily be stored in tanks and loaded onto liquefied natural gas (LNG) tankers for onward transportation.

Figure 1: Conceptual view of the

Prelude FLNG (image courtesy of

Shell International Ltd.)

16

Marine | Simulating a conceptual FLNG unit in waves

To reach the full potential of this technology, it must be ensured that in extremely adverse weather conditions, such as storms and heavy seas:• The ship’s structure is able to withstand

the enormous strains that arise;• And it is possible to maintain operations

with as little disruption as possible, including the docking and loading of the LNG tankers.

In order to gain a detailed knowledge of the conditions to be expected and ensure undisrupted operation, the AMC has analyzed, in a scientific project involving numerical simulation and experimental validation, how such gigantic FLNG facilities behave at sea.

The projectThe three-year research project started in March 2014. The initial phase, which has now been completed, consisted of investigating the influence of different wave frequencies on the motion response of a conceptual FLNG unit.

In the second phase, which is still in progress, the primary focus is on operational aspects of the facility, specifically on the interactions between a FLNG facility and much smaller LNG tankers and supply ships during approach and mooring. These include the emergence of frequencies causing pitching and rolling movements, and undesired resonance waves.

The project is conducted by Yuting Jin, who currently is a Ph.D. candidate at AMC. His intent is to provide specific information to help with the development of the following target areas:• Planning: determine design

configurations suitable for critical conditions;

• Operation: establish efficient procedures for safe operations;

• Crew training: enable precise and practical crew training.

CFD simulations at AMC SearchThe AMC specializes in shipping and maritime engineering. The institute has an extensive range of testing equipment, including a 100-meter towing tank, a circulating water tank, a cavitation tunnel and a 12x35 meter model test basin. Also, it has access to a computing capacity of over 1,500 cores.

AMC Search, the commercial arm of the institute, has been making the acquired knowledge and the developed techniques from research and experimental testing available to the maritime industry in Australia, New Zealand and across the world for over 30 years. Dr. Max Haase is responsible for implementing CFD simulations into commercial projects. He states that in recent years, CFD has played an increasingly important role, due to more sophisticated requirements in performance evaluation and design optimization which cannot be achieved by model testing in a

Figure 2: Model test experiment

in the AMC model test basin

17


timely and cost-effective way. At AMC Search, STAR-CCM+ is a popular choice for CFD studies due to its versatile simulation capabilities, its user-friendliness, and its computational speed.

Towing tank versus simulation Towing tanks have been an indispensable tool for ship design, optimization, and performance assessment for over 150 years. Over time, procedures used have proved their value and achieved a high degree of accuracy. However, model testing is typically not available until a late development stage, when design and construction are well underway. In addition, the construction and alteration of the prescribed scale models can be both time-consuming and expensive. Overall, the flexibility and ability to innovate as required in today’s development cycles is clearly limited by the use of towing tanks only. Furthermore, they are limited to scale models that are significantly smaller when compared to the full-scale device, potentially restricting the ability to investigate innovative designs.

As a result, a growing number of engineers are turning towards numerical simulation in order to assess complex systems at a much earlier stage of the design process. Simulation software, such as STAR-CCM+, has been proven to be as accurate as towing tank tests, and given realistic assumptions, allow ships and offshore platforms to be simulated at full scale,

thereby eliminating some important uncertainties introduced by the scaling process. Scale model testing remains relevant in terms of not only demonstrating software robustness, but also the validity of assumptions relied upon in carrying out various design investigations.

AnalysisThe dimensions of the computational domain for the full-scale calculations were 3,000 x 800 meters. For these calculations, meshes from 4 to 12 million cells were used depending on the wave frequency being investigated. A total of 40 calculations were performed. The calculations required around 700 hours using between 48 and 64 cores. Water depths of between 80 and 800 meters were simulated in order to assess the shallow water effects that may occur during towing tank tests and lead to inaccuracies.

The following STAR-CCM+ features were used:• Overset mesh: The overset mesh

capability permitted easy positioning of the LNG tanker in the vicinity of the FLNG unit, for example to analyze the effects of approach and mooring (for example, resonance waves).

• Motion model: The dynamic fluid-body interaction (DFBI) model was used in order to account for the coupling between waves and ship movement.

• Wave model: The non-linear Stokes 5th order wave model was chosen for its

Figure 3: Analyzing the motion

response of the FLNG unit in

STAR-CCM+

18

Marine | Simulating a conceptual FLNG unit in waves

Figure 4: Resulting waves at

various frequencies

Figure 5: FLNG pitching rate

obtained through both

simulations and experiments,

plotted against the wave

frequency

19


accurate representation of wave propagation in open water. The wave height, set to 4 meters, was determined using BMT Global Wave Statistics for the sea area of interest. Particular attention was paid to wave damping in order to avoid unwanted wave reflection.

• VOF model: The Volume of Fluid (VOF) multiphase model was used in order to correctly capture the interface between water and air, and accurately depict the interaction between the hull and the free surface.

The simulations revealed that:• The wake from the FLNG overlays the

ocean waves and forms a relatively calm area;

• With high frequency ocean waves, steep waves (with deep troughs and sharp crests) are formed around the FLNG.

Future investigations will look at how the berthing of LNG tankers and supply ships will affect this configuration, in particular how to avoid resonance waves between the different hulls, how to control the pitching movement of the ships involved, and whether regulations need to be adopted to make the operation safe. The comparison between simulation and model test results at a model to full-scale

ratio of 1:100 shows an excellent agreement over the entire frequency range. It highlights the impact of a limited water depth especially on the pitching movement of the FLNG for waves of low frequencies (figure 4).

ConclusionThis study highlighted how the use of CFD simulations can help engineers make decisions concerning not only hull design and layout configurations, but also ship operations.

At AMC Search, these results will be used to develop recommendations and operating guidelines for three target areas: planning, operations and crew training. For Haase, this project is valuable for another reason: “Increasing attention is being paid to CFD technology in the hitherto rather conservative maritime field. Nevertheless, compared with Europe where there are a number of model test basins, organizations and service providers with comparable interests, CFD has not yet unfolded its considerable potential for the maritime industry in Australia. We believe that with this project we have demonstrated the capability and scope of CFD simulations and have achieved an important milestone in the establishment of this method for maritime applications.”

Max Haase, Consultant at AMC Search Ltd.Max Haase graduated with a Master of Science in naval architecture from the University of Rostock, Germany, in 2008 and received his PhD in maritime engineering from the Australian Maritime College at the University of Tasmania in 2015. His expertise is the numerical performance prediction of marine surface vessels, ranging from small unmanned autonomously acting rescue vessels to large fast catamarans. His current role at the Australian Maritime College includes academic duties and the implementation of computational fluid dynamics into commercial projects for the maritime industry. Yuting Jin, PhD candidate at AMCYuting Jin completed his Bachelor of Engineering degree in ocean engineering with first class honors from the Australian Maritime College at the University of Tasmania in 2013 and is now pursuing a PhD degree in maritime engineering. His work focuses on computational fluid dynamics for predicting surface vessel maneuvering performance and ship-ship hydrodynamic interactions. He is looking forward to apply the outcome of his research to real-time handling of large offshore structures.

“CFD simulations can help engineers make decisions concerning not only hull design and layout configurations, but also ship operations.”

Acknowledgements:The project has been initiated and supervised by A/Prof. Shuhong Chai, Prof. Neil Bose, Dr. Jonathan Duffy and Dr. Chris Chin at the Australian Maritime College.

20

21

Trends in CFD applications for the maritime industry | Marine

Volker Bertram - DNV GL Maritime Advisory

Hull design is the number one factor in fuel efficiency. It impacts profitability, competitiveness and ship value. Since the first commercial ship basin was commissioned in 1883, towing tanks have provided naval architects with a reliable method of predicting the performance of a ship at sea. Tank testing is commonly used for both resistance and propulsion tests. However, the cost and effort of producing a model and testing it, means that this process is utilized late in the design cycle. This method verifies and fine-tunes an

Trends in CFD applications for the maritime industry

established design, rather than being a tool to help drive and optimize the design.

CFD has long been considered a credible alternative to tank testing. It provides a numerical model that can be implemented much earlier in the design process. Naval architects can make use of engineering data to influence and improve the design process. Another advantage to CFD is the accuracy of results, independent of the scale of calculation.

Figure 1: Wave making before

(top) and after line optimization

(bottom) in STAR-CCM+. This

typically saves 5 percent

compared to simulation-based

approach.

22

Marine | Trends in CFD applications for the maritime industry

prismatic cells to recognize cylinders with extrusion along centerline and thin solids, or gaps, with projection from one side to another. The result being that today, CFD models often provide a higher level of detail than achieved with model tests. CFD software can now handle moving parts (propellers or rudders), model complete systems rather than single parts, and can replace geometry (if required) to perform analysis with and without specific parts.

Turbulence modeling In the 1980s and 1990s unsatisfactory results were often blamed on the limitations of turbulence modelling. This type of modelling is useful for analyzing the flow structures and resulting resistance of bare hulls, as investigated in most validation studies. However, the propeller behind the ship dominates flows and reduces the effect of the turbulence model. For most applications in the marine industry, the standard k-ε or k-ω turbulence models are adequate. But other models are available to better predict secondary flows, the Reynolds-stress model (RSM) currently being one of the most popular options. In the future, large-eddy-simulation (LES) analyses are likely to end the debate on turbulence modelling. LES directly captures the larger, significant ‘finger-print’ vortices of the flow directly and uses subgrid-scale

Preferred approachThe improvements in computing power, have allowed experts working in shipping to use CFD calculations to simulate vessel hydrodynamic performance more accurately and faster than ever before. The industry’s ability to handle complex geometry with all relevant details has also greatly improved. Development in grid generation has made it easier to generate high-quality grids for accurate CFD simulations.

Many aspects have advanced the wide acceptance of CFD as a design and optimization tool. The increase in hardware power combined with progress in various aspects of the flow solvers permit a wider scope of more sophisticated applications. Such analyses have become increasingly important and have now resulted in CFD surpassing model tests as the preferred approach for many applications in the maritime industry.

Developing techniquesMore sophisticated CFD analyses for ships and offshore platforms employ a variety of techniques that have become widely available in recent years. One key aspect for carrying out calculations based on complex geometries, such as analysis of offshore platforms, is geometry recognition. In this case the pre-processing software uses

Figure 2: Model of a complete

propulsion system in STAR-CCM+

23

Trends in CFD applications for the maritime industry | Marine

turbulence models for the small, “background noise” turbulence. Currently only a few research institutions have the computational resources necessary to carry out LES calculations. However, these resources are expected to become available to the industry over the next ten to fifteen years through a general growth of computing power and cloud-based business models.

Classification society approvedFree-surface flows are of great interest to naval architects. Measuring the wave resistance of a ship can help them determine which small or moderate changes in hull shape could significantly reduce the overall resistance of the vessel and improve its performance. Other applications of free-surface flows include seakeeping, slamming and sloshing. Modern CFD methods allow the simulation of highly nonlinear free surface flows. Such simulations are now so well predicted that they are widely accepted by classification societies for load determination in strength analyses.

Easier to useCFD tools have become more user-friendly as reflected in the use of integrated design environments. The integrated design environment combines many aspects of CFD software including free-form hull description using parametric modelling, interfaces to most

modern CFD solvers, several optimization algorithms, and software to handle process management and user interfaces. The design engineer can then work on simulation driven designs using one interface from model generation to post-processing.

Growing computer power and fully automatic procedures have opened the door for formal optimization as the natural step beyond simulation-based design. Lines optimization (also local bow optimization for refits in times of slow-steaming) saves typically 5 percent beyond the simulation-based approach. Trim optimization saves typically 3 percent beyond the classical approach based on crew experience.

Leave it to the expertsDespite the growing power of CFD software, it remains a tool. The speed and quality of results achieved depends on the person using the tool. Effective CFD results are achieved through a combination of knowledge, understanding and skillful CFD techniques. Despite progress in number crunching, expertise and competence remain at the core of good engineering.

CreditsThis article first appeared in ShipBuilding Industry, Vol. 9 Issue 5. Many thanks to Yellow & Finch Publishers – ShipBuilding Industry, for allowing us to reuse it.

Figure 3: Level of detail of CFD

grid for complete oil rig (left) and

simulated air flow field (right) in

STAR-CCM+

Figure 4: Modern CFD methods,

such as STAR-CCM+, capture

highly complex free surfaces

such as sloshing analyses

24

25

Alessandro Fiumara - Assystem France and ISAE-SupaéroJulien Senter - Assystem France Nicolas Gourdain and Vincent Chapin - ISAE-SupaéroPrashanth S. Shankara - Siemens PLM Software

The America’s Cup, affectionately known as the “Auld Mug,” is the oldest sporting trophy in the world dating back to 1851. Pushing the limits of sailing and sailors alike, the prestigious competition has seen a surge of interest in recent years riding on the coattails of fast yachts, technological innovations, high profile athletes and teams backed by billionaires. Catamarans introduced in the 2013 America’s Cup were faster than the wind, thanks to the introduction of hydrofoils and of solid two elements wingsails, which share similarities with high-lift aircraft wings. The wingsails were based on rules and guidelines outlined in the AC72 class, used in the main race, and the AC45 class, used for preliminary races and training. The newer class catamarans are unlike any other previous catamarans, reaching twice the wind speed and hydrofoiling on water, leading to a surge in spectator interest. The 35th America’s Cup will be held in June 2017 on the Great Sound of Bermuda and will be raced in the new AC50 class, a wing sail powered, fast, foiling catamaran smaller than the AC 72, manned by a six member crew.

Compared to conventional “soft” sails, a wingsail is much more complex, providing lift with variable camber, controlled by a flexible or jointed structure. The wingsails offer greater aerodynamic efficiency

Know the ropesAssystem Investigates the America’s Cup Wing Sails

compared to the canonical sails and better performance, as seen by top speeds of around 47 kts (87 km/h) in the races. While this has made the races faster and more exciting to the viewing public, the challenges lie - as always - in handling these wingsails and managing their aeroelastic behavior to achieve the highest performance during navigation.

These wing sails are difficult to control and the research on their stability in multiple scenarios is still evolving. Finding a stable setting in all navigation conditions for these wing sails is challenging and there have already been a couple of instances of catamarans “flipping” or “falling over” due to this instability. As in the aerodynamic stall of an airplane wing, these wing sails can stall during operation and it is crucial for designers to understand their stall behavior under different wind conditions.

To further the understanding of wing sail behavior, a PhD research project was initiated by Assystem France, an international engineering and innovation consultancy, and the Department of Aerodynamics, Energetics and Propulsion (DAEP) at ISAE-Supaéro, the renowned French Aerospace Engineering School.

The wingsail aerodynamic behavior near the stall conditions is not well understood

Know the ropes | Marine

26

as well as the sailing envelope of this rig still largely unknown. Thus one of the goals of this work was to describe and better understand the flow characteristics on the wing sail in order to identify the relation between design and trim parameters and performance to find the optimum wingsail configuration for each operating condition.

Choosing the analysis methodTo properly investigate the wing sail behavior, the researchers designed a wing sail based on the AC72 class catamaran of the 2013 America’s Cup. Wind tunnel testing is the first and foremost option but is limited by the number of tests that can be run, cost of operation and does not lend itself to easy parametric analysis and design optimization. As with all the marine designs in this day and age, the team decided to use numerical simulations to study the wing sail behavior. In addition to being cost effective and suitable for design sweeps and

optimization, a computational fluid dynamics (CFD) analysis can shed light on some of the characteristics difficult to investigate in a wind tunnel.

The next question is the choice of tool for the numerical simulation. STAR-CCM+® Software, already widely used at Assystem France, was selected to perform these analyses. Aside from convenience, the choice of STAR-CCM+ was based on:• High quality polyhedral meshing that

allowed excellent capture of wake flow;• Flexibility of STAR-CCM+ to be tailored to

a specific problem and option for design optimization;

• Excellent academic licensing options which enabled multiple simulations to be run at ISAE-Supaéro;

• Well established in the marine industry.

Wind tunnel tests modelingValidation of the numerical method with experimental data is the key to accurate analysis. A 1/20th scale model of the wing sail without the catamaran was built for wind tunnel testing in the S4 wind tunnel (figure 2) facility at ISAE in Toulouse. The wind tunnel is an Eiffel (open return) wind tunnel, with an open test section of elliptical shape. The geometry is a two-element swept wing with a main element and a flap with a constant slot between elements along the wingspan. The flap angle can be modified by deflections to change the global camber of the wing.

Wind tunnel tests were performed both for the numerical validation of STAR-CCM+ at different wing sail settings and to obtain an exhaustive experimental database for the flow analysis. The aim of the research project is not directly to find the best setting during navigation but to completely investigate the flow mechanisms around the wing sail, deepening the analysis on the role of the jet flowing in the slot dividing the two wing sail elements, in the enhancement of the high-lift capabilities.

The numerical analysis in STAR-CCM+ was conducted on the same wingsail geometry reproducing also the wind tunnel domain in a way to take into account the interference between the wing and the facility domain [1]. The flow condition in the wind tunnel reproduced the experimental test characteristics both in velocity and in turbulence intensity. The computational domain was discretized with polyhedral cells in STAR-CCM+ (figure 3).

Figure 1: Assystem America’s Cup

yacht type model with wing sail

Marine | Know the ropes

27

Prism layers were automatically built at the surface to accurately resolve the boundary layer flow. The final mesh count was approximately 32 million cells. Unsteady RANS (URANS) equations were solved with incompressible fluid. The flow speed in the wind tunnel duct was 20m/s resulting in mean chord Reynolds number of 5.3x105. These flow conditions with moderate Reynolds number was challenging because they may give rise to laminar flow regions on the wingsail surface. The transitional k-ω SST turbulence model in STAR-CCM+ was used to account for the boundary layer transition from laminar to turbulent, predicting flow separation.

The simulations showed excellent comparison with the wind tunnel data, both quantitatively and qualitatively. A comparison of flow separation on the wing between the wind tunnel test data and the URANS simulation shows that the flow characteristics are similar in the two approaches, with the flow separating on the higher sections of the flap while staying attached on the lower sections.

Similar to the qualitative results, the quantitative results from the STAR-CCM+ simulations agreed well with the test data, reinforcing the validity and accuracy of the research method. A comparison of the pressure coefficient on the main element in a large slot configuration for high and low camber is shown in figure 6. In areas of

attached flow, the numerical and experimental data match very well for this configuration. In areas of separation on the flap surface, the numerical results are within 10 percent of the experimental data. This process has helped to validate the numerical methodology to predict the performance of the wing sail in different conditions [2].

From the simulation, the low cambered setting appears to be the best option for an upwind condition while for the downwind condition, the higher flap deflection angle

Figure 2: Test model of wing sail

in S4 wind tunnel (left)

Figure 3: Polyhedral STAR-CCM+

mesh around the wing sail (right)

Figure 4: Streamlines colored by

velocity around the wing sail,

showing skin friction coefficient

on the surface


28

performs better. The high camber configuration shows a certain sensitivity to the slot size dividing the two wing elements; a small modification of this size can offer performance improvements.

Results show the high sensitivity of the stall behavior to the flap deflection angle and slot size [3]. In certain wingsail configurations, the wing has a multi-step stall with sections of the flap stalling at different angles of attack, while the main element stalls for the highest angles of attack [4]. Wind tunnel tests and URANS simulations agree to show how a modification of this slot size will eliminate the multi-step behavior while offering a linear increase in lift coefficient. The same slot size can change the final stall on the main element from abrupt to flat stall. Understanding and achieving a right slot size along the entire wingspan is of crucial importance for operational stability during the race but challenging because of the aeroelastic behavior of the wingsail in real conditions.

Modeling of a C-Class catamaranTo fully analyze the wing sail behavior in downwind condition, the geometry was scaled to a C-Class catamaran dimension. C-Class catamarans are used in the LITTLECUP race, the testing ground for the America’s Cup, where the latest technology in foiling catamarans is tested, compared and approved. These are also used in the International C-Class Catamaran Challenge Cup, an open design class with very simple design rules. The C-Class catamarans are known for their simplicity with a sail area that must not exceed 27.8 square meters, forcing designers to push the limits of design innovation to go faster.

The scaled down wing sail geometry was modeled along with the catamaran geometry to consider the wing-catamaran aerodynamic interaction in downwind condition. The sea boundary layer was modeled including wind twist, with lower angle of attack on the low sections, and higher angle of attack upward

on the sail. The flap deflection angle was set to 35°, closer to the angle set by sailors in downwind conditions. In addition, flap deflection angles of 15° and 25° were also considered.

Figure 7 shows the comparison of pressure distribution (Cp) for the different flap deflection angles. Contrary to conventional wisdom, the 35° flap deflection angle was not found to be the angle that maximizes the driving force in downwind conditions. The driving force and heeling moment were calculated for the different angles, with the best performance achieved at maximum driving force and minimum heeling moment. The results show that at the conventional 35° angle, the flow is completely separated on the flap surface while it remains attached at 25°, increasing the driving force on the wing.

The wing sail response to gusts was also researched by introducing a sinusoidal law at the inlet velocity in the simulations. As seen in figure 9, “mushroom” separation cells appear on the flap surface and spread with a wave structure to the entire wing sail span under gusts. It will be interesting to know if these flow patterns may be observed in real sailing conditions.

ConclusionThis research has laid the groundwork for understanding the stability of the wing sails in the America’s Cup races and LITTLE CUP to be able to design better, faster and more stable wings for foiling boats. Future research will include the design and trim optimization to achieve the best driving force under a constraint of the maximum heeling moment bearable by the catamaran. Also, the origin and influence of the “mushroom” cells need to be investigated. The research team is also looking to collaborate with a sailing team from the 35th America’s Cup as a design partner. The use of numerical simulation has and will keep pushing the limits of speed and technology in the America’s Cup races.

Figure 5: Skin friction coefficient

comparison on wing sail from

viscous oil visualization in wind

tunnel (left) and STAR-CCM+

(right)

Marine | Know the ropes

29

References[1] Fiumara, A., Gourdain, N., Chapin, V., Senter, J. (2015), “Comparison of Wind Tunnel and Freestream Conditions on the Numerical Predictions of a Flow in a Two Element Wingsail”, 50th AAAF International Conference, Toulouse, 30 March – 1st April 2015. [2] Fiumara, A., Gourdain, N., Chapin, V., Senter, J. (2016), “Numerical and Experimental Analysis of the Flow around a Two-Element Wingsail at Low Reynolds Number”, submitted in 2016 to the International Journal of Heat and Fluid Flow. [3] Fiumara, A., Gourdain, N., Chapin, V., Senter, J. (2016), “Aerodynamics Analysis of 3D Multi-Elements Wings: an Application to Wingsails of Flying Boats”, RAeS Applied Aerodynamics Conference, Bristol (UK), 19th-21st July 2016. [4] Chapin, V., Gourdain, N., Verdin, N., Fiumara, A., Senter, J. (2015), “Aerodynamic Study of a Two-Elements Wingsail for High Performance Multihull Yachts”, 5th High Performance Yacht Design Conference, Auckland, 10-12 March, 2015.

Figure 6: Comparison of Cp for

low and high camber

configurations between wind

tunnel and STAR-CCM+ (top left)

Figure 7: Cp comparison along

the wing for different flap

deflection angles (top right)

Figure 8: Pressure coefficient

on wing sail and flow

streamlines (left) and velocity

magnitude at 25 percent, 50

percent and 75 percent of

wing span (right)

Figure 9: Skin friction

coefficient on wing showing

“mushroom” separation cells

(below)


30

31

Kuninori Masushige and Yuka Takahashi - Siemens PLM Software

Taking full advantage of computational fluid dynamics (CFD) in electrical and thermal energy technology innovation, seasoned engineers from the Thermal System Technology Department in the Application Technology Research Center at Fuji Electric Co., Ltd. (hereinafter, “Fuji Electric”) are applying computer-aided engineering (CAE) to the performance reviews and design development technology of various products developed by the company. In this article, Fuji Electric simulation experts Tsutomu Yamamoto, Yoshiaki Enami, and Kimihisa Kaneko discuss how they apply CFD.

Established as a capital and technology alliance between Furukawa Electric Co., Ltd of Japan and Siemens AG of Germany, the name “Fuji” is derived from the first syllables of the two company names, “Fu” of Furukawa and “Si” (according to Japanese phonetic pronunciation) of Siemens. Fuji Electric, with over 90 years of electronics manufacturing, has expanded into a broad range of fields, including power generation and social infrastructure, industrial infrastructure, power electronics, electronic devices, and food distribution.

Fuji Electric’s analysis consulting unitYamamoto’s group belongs to the Advanced Technology Laboratory, where research and development into the common infrastructure and advanced technology of the entire Fuji Electric company is conducted. As a department that regards CAE as its core technology, it tackles the development of new technology related to areas such as structural, thermal-fluid, and electromagnetic field analysis, while at the

Contributing to the efficient use of electricity and heat in power generation

same time collaborating and supporting in-house product development and design. As Fuji Electric’s products span a broad range, including power generation plants, substation equipment, power electronic devices, power semiconductors, and vending machines, this department conducts its operations in cooperation with each division with a focus on analysis evaluation.

The group receives analysis requests from each division in the company. These include not only straightforward simulations but also cases relating to improvement proposals to deal with various problems in product development, such as what type of structure can solve the problems at hand or how performance and reliability can be enhanced. When dealing with these requests, they are conscious about consulting and not just performing analysis.

We spoke to the three people in charge of simulation at Fuji:

Yamamoto: “I am currently managing this division. A variety of requests come from inside the company, so we act as a bridge between our department and others. I also assist in the direction of all divisions that focus on CAE, and I plan and substantiate research and development goals. Previously I carried out thermal fluid analyses for cooling studies in power generation equipment and power electronics, however I still partially perform analysis work.”

Enami: “I specialize in thermal fluid analysis. Some of our work involves electromagnetic fields, such as investigating

Contributing to the efficient use of electricity and heat in power generation | Electronics

32

arc1 phenomena, so I also conduct electromagnetic field analyses. In addition, for the past year or two, I have been working on the development of a new technology for applying geometric optimization using the adjoint state method.”

Kaneko: “My responsibility lies mainly in cooling studies for power electronics. For the most part, we provide developmental support for inverters and motors.”

Products and technology that have risen from interdisciplinary projectsThis 10-person group at the Advanced Technology Laboratory is primarily responsible for CAE, with the main fields being structural and thermal fluid analyses to evaluate product performance and soundness. As specialists in CAE, they develop coupled simulations to conduct design analyses in fields where this was not previously possible. A recent, particularly hot topic is the relationship between arc behavior and the current interruption obtained by coupling thermal, fluid and electromagnetic field analyses.

Other divisions of the laboratory, as well as the design divisions of each branch, also carry out various simulations and they cooperate with each other. When conducting joint development of new technology with different design divisions, they perform design evaluations at an early stage. Projects are transferred sequentially to each branch when the technical development of the simulation is complete and when an evaluation method has been established.One of the strengths is that the group is

involved in a wide range of fields. When they look at the core technology behind each product, they ask: “Can we use this technology for other products?” For example, part of the heat transfer enhancement technology in product cooling has been deployed in film formation and cleaning for semiconductors, and analogies between heat and mass transfer can be applied to other areas.

CFD: An indispensable tool in product developmentYamamoto discussed the necessity of CFD and the difficulties in its application: ”In particular, the implementation of prototype tests with plant equipment for power generation and substations is difficult. Detailed measurements are difficult in the first place and without CFD new product development would be difficult. Also, in the case of power electronics, very short development periods are required,” Yamamoto says. “Product development is conducted through a combination of prototype evaluation and verification by analysis, but detailed measurement and evaluation cannot be conducted unless the housing design is at a certain degree of advancement, particularly for cooling. A shorter development period can be achieved by conducting analysis in advance and by narrowing down the proposed structural design.”

Kaneko adds, “The difficulty for me is that it is not easy to standardize or routinize the evaluation and investigation methods because there is such a variety of

Figure 1: Thermal System

Technology Department,

Application Technology Research

Center, Advanced Technology

Laboratory, Fuji Electric Co., Ltd.

Group Manager Tsutomu

Yamamoto (left), Yoshiaki Enami

(middle), Kimihisa Kaneko (right)

Electronics | Contributing to the efficient use of electricity and heat in power generation

33

products. If the product structure has been decided to a certain extent or if evaluation for a case study is conducted, we can automate these methods using macro programing so as to greatly improve work efficiency. However, the difficulty in our division is that the product range is too large.”

Products subjected to the group’s analysis are often bespoke rather than mass-produced products, and for this reason they have to start from the geometric creation of the model. When viewed in this light, the STAR-CCM+® Software mesher is a very powerful tool because it can deal with a series of process steps in a short time period after receiving 3D CAD from the relevant divisions of the company. The group says that they receive huge benefits from STAR-CCM+ as they are able to quickly move from mesh creation to performance evaluation.

“The calculation of the current in the arc analysis is carried out with the electrodynamic potential solver in STAR-CCM+, in a double-precision version. This is because when you set the original value of the conductivity of the metal in a mixed-precision version, the simulation is likely to diverge. As I understand it, currently a finite element method (FEM) solver is being developed and we expect that it will resolve the issue of not being able to maintain current continuity in parts where plasma conductivity changes abruptly. If electromagnetic field analysis can be implemented with FEM, I’d definitely like to try it,” says Enami.

Reducing the turnaround time: Simplification of models based on the engineer’s knowledgeIf detailed models are created for cooling studies of products with complicated structures such as generators or motors, or for temperature evaluation of printed circuit boards or power electronics that have a large number of components, they become very large-scale. As there are also limitations in hardware, they take the approach of narrowing the focus point while combining CFD and thermal network calculations on sub-models.

Current resources are sufficient in RANS calculations. However, when modeling complex structures the mesh increases considerably. Thus, it is necessary for engineers to consider this and proceed with simplification.

“It is possible to incorporate the entire CAD geometry as-is into the CFD model, however, the time required for meshing and post-processing also increases. When looking at results with regard to geometry, we select the most appropriate method for individual cases based on what level of precision is required, what the acceptable analysis period is and how accurate the measured data is,” says Kaneko.At the present time, conducting analysis after a certain degree of geometric simplification is determined to be faster because the comparisons of multiple conditions are often required.Kaneko says, “In the future, computers will become even faster and abundant resources will become available, and if calculation technology evolves alongside this, the situation will change.”

Development of a unique method in arc analysisWhen asked about arc analysis as an example, Enami said, “When the arc analysis was first developed, it started from an arc between simple rod-shaped electrodes, but now it is possible to predict the arc voltage applied to products such as molded case circuit breakers, air circuit breakers, and circuit protectors.”

The group struggled to stabilize calculations in the development process. For this reason, a magnetic field calculation function based on the Biot-Savart Law2 was developed by themselves independently. While stability is very high with this

Figure 2: Arc analysis flow chart


34

method, there is the problem of an increase in calculation time in the N2 order, and this is accelerated by using a parallel calculation using a GPU.

In addition, for magnetic materials such as iron, the magnetizing current is worked out and the magnetic field is calculated using the surface current method – a type of boundary element method. A stable solution can be obtained although there are some limitations such as the inability to take into account the B-H property3 and the eddy current4 as it is a linear analysis, along with the inability to easily increase the number of surface elements, because a direct method is used by LU decomposition5.When the geometry of the magnetic material is complicated, or when there are large numbers of parts, large divergences with experimental results tend to occur. This highlights the limitations of the surface current method. Accordingly, there are high expectations for magnetic field calculations with STAR-CCM+ FEM.

Considerations when developing global productsIt is said that products for overseas markets often demand higher environmental performance standards compared to products for the domestic Japanese market. In particular, electrical equipment, for

railways or plants exposed to outdoor environments. So a structure that does not allow sand, dust, or snow to enter the intake unit becomes a priority, thus investigations with Lagrangian two-phase flow models are conducted. In the case where filters are attached as a measure against sand and dust, it is required that no problems exist with the specifications of the internal temperature of the equipment. In addition, fuel cell products for the European market require compliance with standards such as CE marking6. In order to meet explosion proofing standards, CFD is used to confirm that flammable gas concentrations, at the time of potential leakage, are less than the explosion limit.

Information-sharing of CAE within Fuji ElectricAs a cross-sectional position in the company, an organization called the “Design Technology Group” has been established and it strives to strengthen common basic technologies involved in design, such as 3D-CAD design and EMC7 measures. Also, an organization called the “CAE Liaison Committee” has been established within this group. Liaison meetings, in which key CAE personnel from each department participate, are held four times a year. Information relating to current hot topics and technical problems

Figure 3: Arc analysis (circuit

breaker for wiring)

Figure 4: Arc analysis (air circuit

breaker)

Figure 5: Arc analysis (circuit

protector)

Figure 6: Circuit protector arc

voltage

Electronics | Contributing to the efficient use of electricity and heat in power generation

35

are shared, and CAE-related training sessions within the company are planned; Yamamoto is the leader of this committee.

The products covered by the group span a wide range of target products and fields, from semiconductors to generators. For this reason, the person in charge of each project may have some deficiencies in terms of knowledge and technology, but on the other hand the strength is that there are a number of members with experience in a variety of products and fields.

If CFD is applied to product development, it is necessary to analyze and assess the ana-lytical errors and areas of improvement for the product after looking at the simulation results and measured data. In that context, junior group members tackle not only CAE but also actual measurement verification. In addition, training is also undertaken to learn the operation of STAR-CCM+ and STAR-CD® Software.

Responding to sudden requests, a skill that is cultivated in the company“Our task inevitably involves sudden requests and in order to respond quickly to these, we must listen to those making the request and to product developers with regard to the design intent and performance of conventional products,” Yamamoto says. “When there is physical testing data available on similar existing products, we consider this to determine our simulation strategy. In addition to CAE, the skill to conduct the comprehensive evaluation of product performance is important.”

Reasons and requirements for using STAR-CCM+STAR-CCM+ is used almost every day as a main work tool and is extremely helpful. The group is satisfied with its considerably extensive features. The automation by Java macros is easy to use and in this respect, STAR-CCM+ has much more flexibility than STAR-CD. The STAR-CCM+ Optimate+ add-on is also available in the workplace and the group plans to try parametric optimization in the future.

The group also thinks that it is very helpful that support and services are always efficient and effective in helping them solve their engineering challenges. The company received the reputation that its best practices manuals and support

materials were substantial, but at the same time there was a request for continued expansion. Best practices for analysis procedures and methods for setting conditions that become basic in various different fields are very useful because the group is often asked to make evaluations in relatively short periods of time by various departments at Fuji Electric.

ConclusionBased on an in-depth knowledge of CAE and of CFD gained over 20 years using STAR-CD and STAR-CCM+, these three individuals incorporate engineering knowledge into analysis and conduct simulations with a very high degree of difficulty. The development of their own approach, such as the simplification and coupling analysis of models based on engineering skill in order to shorten the turnaround time, was very impressive.

Figure 7: Fuel cell analysis

(temperature distribution)

*1 “Arc” is a phenomenon in which hot gas becomes conductive by being ionized at temperatures from several thousand degrees to tens of thousands of degrees. Heat is generated by the current, causing a rise in temperature, which maintains the current flow. The prediction and control of the behavior of the arc at the time of the opening and closing of the contacts in the components of power reception/distribution devices and control equipment is an important technical issue.*2 The Biot-Savart Law is a law in electromagnetism for calculating the magnetic field being generated around a current.*3 The B-H property is a curve showing the magnetization process of a magnetic material, in which changes in magnetic flux density B is plotted against external magnetic field H.*4 An eddy current is a spiral current that occurs in the conductor due to electromagnetic induction.*5 LU decomposition is a matrix decomposition method in mathematics, used to solve simultaneous linear equations.*6 CE marking is a symbol indicating compliance with standards and is affixed to specified products for sale in the EU. *7 Electromagnetic compatibility (EMC) refers to the situation when electrical and electronic equipment do not cause electromagnetic interference that may affect the device itself or other equipment.


36

37

Gaetan Damblanc – Siemens PLM Software

Figure 1: “Build and Break”

conventional process to meet

cell’s requirements as seen on

the spider chart

IntroductionThe use of Lithium ion (“Li-ion”) batteries is ubiquitous: they are used in phones, cameras, laptops, cars, watches, and more recently, hover boards. They are also used in larger systems, such as ships and airplanes. The demand for safe and high performance Li-ion batteries has never been higher.

Cell and battery pack manufacturers are continuously seeking higher energy density (or specific energy), higher power density (or specific power), safer products, longer lifespan and lower cost. Significantly improving a battery design across its whole operating range is a challenging task and involves the simultaneous optimization of numerous parameters. Large cell and pack producers know this very well: it is both cost and time-consuming to test and validate all the different material combinations. Accelerating the design process while reducing costs is an objective battery manufacturers are obsessed with and

Driving battery innovation with CAE

computer aided engineering (CAE) is part of the solution.

At the cell level, Battery Design Studio® (BDS), a powerful cell design software and cell testing platform, allows multiple cell designs to be assessed at a fraction of the cost and time usually required for experimental work. At the pack level, STAR-CCM+® Battery Simulation Module (BSM) predicts the complex electro-thermal behavior of the whole pack with high accuracy, a critical component for xEV powertrain design.

This way, hundreds of designs can be tested at low cost. However, virtually building each cell/module takes time and requires continuous analysis for the design to be improved. This can be done automatically by coupling BDS and BSM to design exploration and optimizing software, such as HEEDS™.

This article will demonstrate how a well developed, commercially available cell can

Driving battery innovation with CAE | Ground Transportation

38

still be improved by coupling BDS and HEEDS.

Cell descriptionThe commercial cell considered in this study is a cylindrical cell of Type 36650, which means it measures 3.6 cm in diameter and 6.50 cm in height. It is a high power cell and the chemistry used for the cathode is a blend of Nickel Cobalt Manganese (80 percent) and Lithium Manganese Oxide (20 percent). The anode is made of graphite.

Because it is a high power cell, it can operate at very high current while still retaining most of its available energy. Typically, the power capability and the available energy are referred to as power density and energy density. These are the power and energy available per mass unit and are expressed in W/kg and Wh/kg, respectively. Battery cells are systems that cannot provide high power density with high energy density, one of the trade-offs in cell design. However, it is still possible to optimize the amount of energy such a high

power cell can contain, which is the objective in this example.

Before starting the optimization, it is important to accurately characterize the reference cell in BDS, for example, specify the geometrical dimensions of each part, such as electrodes, coating, tabs, etc., as well as the physics-based performance model in order to predict the cell behavior. Additionally, tab design, electrode dimensions and coating formulations can be easily input in BDS’ user-friendly interface.

Optimization 1: Energy densityWith this reference cell built, the optimization work can start. Since the objective is to maximize the energy density (Wh/kg), the changes will be focused towards weight reduction and increasing the coating length in order to add more active material in the cell and therefore more energy.The following design variables were selected for the study:• Positive electrode: Length, number of

tabs and current collector thickness• Negative electrode: Length, number of

tabs and current collector thickness• Positive tabs: Width• Negative tabs: WidthEach of these design variables evolve within relevant constraints so that they make both physical and manufacturing sense. The design exploration study will perform an analysis of 100 designs.The results show a significant energy density increase, approximately 60 percent compared to the reference case (figure 5).

HEEDS has useful outputs to easily visualize the different parameter changes and highlight combination trends which give best results. This can be seen on a “Parallel Plot” like in figure 6. Highlighted by the green curve are the designs that achieve the highest energy density. The best results are achieved with low current collector thickness (Neg_CC_T, Pos_CC_T) and a high tab count (Pos_Tab_Num, Neg_Tab_Num). The yellow curve shows the combination for the best design, where it can be seen that energy density is much higher than that of the reference design highlighted in gray.

However, it can be seen that the best design has a higher material cost than the reference design. It would be ideal to have the best of both worlds, which is increasing the energy while reducing the cell material

Figure 2: Cell being dissected to

study tabs and electrodes

designs

Figure 3: BDS user interface

Ground Transportation | Driving battery innovation with CAE

39

Figure 4: Cell voltage evolution

during 1C discharge: comparison

between simulation (solid) and

experiment (dots)

Figure 5: Performance (energy

density) for each design

“The ability to automate the simulation set-up and

computations brings a significant gain in

productivity; the entire cell design exploration

study was conducted in one day.”

Driving battery innovation with CAE | Ground Transportation

40

price at the same time. This is the objective of the next optimization study.

Optimization 2: Material cost versus energy densityFor the second study, in addition to the inputs of the previous study, the material cost is also incorporated, something which can be handled by BDS. The design exploration study becomes a multi-objective study in which energy density will be maximized while cost is minimized. The optimal result will not be one single best design but a set of best designs. This will yield a Pareto front showing a trade-off of optimum designs between energy density and cost.

From this Pareto front two designs were selected (out of nine) that match both requirements in terms of energy density and cost as seen in table 1.

Designs 85 and 119 both show increases in energy density compared to the baseline design. Design 85 is closest to the best design in the previous study in terms of energy density but shows a $0.09 material cost reduction. It may appear as a small improvement, but when multiplied by hundreds of thousands or millions of cells produced, this has an impact. Alternatively, if one is looking at higher cost savings, design 119 is a good choice with $0.17 cost

reduction, but still offering a 31 percent energy density increase.

ConclusionImprovements into modeling of Li-ion cell behavior have made associated CAE a powerful design tool. It allows for a tight coupling between the electrochemical and thermal problem which provides great accuracy in predicting these complex systems.The ability to automate the simulation set-up and computations brings a significant gain in productivity; the entire cell design exploration study was conducted in one day. Design exploration using BDS and HEEDS enables optimized cell performance by incorporating both physical and cost performance objective analysis, resulting in better designs, faster based on a number of related parameters that would be time-consuming to achieve in a manual approach. This demonstrates that CAE software is a powerful tool to design and size cells and packs, not only because of the tight coupling between the electrochemical and thermal aspects of the problem at hand in these complex systems, but in the way design exploration studies can be used to rapidly analyze the design space.

Figure 6: Parallel plot - The three

left parameters are inputs, the

rest are responses. Gray is

reference design, yellow is best

design (top left)

Figure 7: Pareto front plot - This

graph plots the set of optimum

designs as a tradeoff between

energy density and cost. The

black line serves only as an

indicator to show the Pareto

front. The gray square is the

reference design. Circled in blue

are designs 85 and 119 discussed

here (top right)

Table 1: Best designs from Pareto

front analysis

Ground Transportation | Driving battery innovation with CAE

DesignsEnergy density improvements

(%)

Cost ($)

Capacity (Ah)

Pos_CC_T (mm)

Neg_CC_T (mm)

Neg_Tab_W (mm)

Pos_Tab_W (mm)

Pos_Tab_Num

Neg_Tab_Num

Baseline - 4.49 7.71 24 16 6 6 4 4

85 60 4.4 7.71 30 10 5.4 8.4 24 15

119 31 4.32 7.79 28.7 8 5.4 8.4 24 5

41



siemens.com/mdx

42

43

Ford Otosan improves diesel engine design | Ground Transportation

Sinan Eroglu and Serdar Güryuva - Ford OtosanChristopher Beves - Siemens PLM Software

IntroductionThe design of the next generation of highly efficient diesel engines is reliant on rapid, robust and reliable performance predictions. This has led the Powertrain CAD/CAE team at Ford Otomotiv Sanayi A.S. (Ford Otosan) to use STAR-CCM+® Software to analyze the thermal exchange between the hot engine gases and various critical solid engine components. Ford Otosan was founded in 1959 as a joint venture between the Ford Motor Company and Koc Holding and is based in Turkey with five facilities employing 12,000 people producing the Ford Transit, the Transit/Tourneo Courier and Ford Cargo trucks.

The use of simulation driven development and design has been crucial to the success of their recently launched heavy duty engine, the Ecotorq 13L/9L, with EU3, EU5 and EU6 variants. The cost and time needed to manufacture prototypes for testing so early in the design phase before more mature designs are ready is high. This sees the increased use of simulation as early in the design process as possible.

Modern diesel engines are complex assemblies with tight packaging constraints and, due to the higher temperatures as a result of turbocharging, have largely non-uniform temperature distributions which are challenging to predict. As a result, the need for increased simulation accuracy has seen the rise of multi-component, multi-physics computational fluid dynamics (CFD) simulation as a fundamental requirement inherent in modern diesel engine design. To this end, the following work at Ford Otosan is presented on two critical engine components: the engine exhaust manifold and the piston head, to see how the digital

Ford Otosan improves diesel engine design with STAR-CCM+

twin was deployed to help them achieve their design goals by the use of co-simulation. As Sinan Eroglu, the CFD supervisor of the Powertrain CAD/CAE group, explains, “STAR-CCM+ co-simulation enables coupling of multi-physics simulations with different time scales ranging from microseconds to thousands of seconds, providing faster and more accurate analyses and shorter turnover times for development and assessment of complex designs.”

Engine exhaust manifold simulationThe function of the exhaust manifold is to bridge the gap between the engine structure and the exhaust after treatment systems and allow the burnt in-cylinder gases to flow through it. Exhaust manifolds are designed and developed to provide smooth flow with low back pressure, whilst being able to withstand extreme temperatures. As Sinan Eroglu explains, “The aim is to come up with a durable manifold design, so accurate modeling of the warm up time is extremely important. It is a typical multi-physics problem where there is a strong interaction between the fluid and solid domains, so the solvers for each of these have to be coupled. The main challenge arises from the nature of the pulsating flow behavior within the manifold, requiring a transient analysis to be conducted. The time step of this flow is in the order of 10 microseconds, and as the warm up period is 10 minutes, it is not feasible to run with this conjugate approach, so we sought alternative co-simulation methods.” For the exhaust manifold, three approaches, shown in figure 2, were assessed.

44

Ground Transportation | Ford Otosan improves diesel engine design

Sequential coupling methodThe first approach, which sequentially couples the finite volume solver in STAR-CCM+ to the finite element solver in Abaqus, is driven by exchanging data between the two codes at the interface boundary of the fluid and solid domains. This allows a more complete structural and thermal stress analysis to be carried out on the manifold; the two models are seen in figure 3. The data exchange is handled by internally

developed Java scripts to conduct the process automatically. Each CFD run is conducted at the rated power condition for three engine cycles (2160° crank angle) with boundary conditions at the inlets and outlets provided from one-dimensional engine performance numerical analysis. At the end of the third CFD engine cycle, thermal load data (heat transfer coefficients and reference

temperatures) are time averaged and mapped onto the finite element model, which is then run for 600 seconds. This data is then fed back into the CFD model, updating the thermal distribution at the interface boundary, and the simulation is continued. Because of the data exchange between the fluid and solid models, the primary concern is temperature convergence, or at what point each “separate” fluid and solid model is up-to-date and there is conservation of energy throughout the whole system. This occurred after the third data exchange, as shown in figure 4.

Co-simulation methodFor cases where a full thermal stress analysis is not required, a STAR-CCM+ to STAR-CCM+ co-simulation can be utilized: here, the finite volume method is used for both fluid and solid models to predict the thermal distribution. This allows the data transfer to be handled solely within STAR-CCM+, permitting a more direct data transfer whilst the simulation runs rather than transferring the averaged thermal load at the end of a complete cycle as is the case with the STAR-CCM+ to Abaqus approach. In this instance, the data was transferred every five “fluid” time steps. Different time steps were used in the fluid and solid models, of 10 microseconds and 0.1125 seconds respectively. 5,333 data exchanges between the fluid and solid models were made in order to ensure convergence.

The conjugate heat transfer methodConjugate heat transfer (CHT) simulations are the most direct approach, where the fluid and solid models are solved within a single STAR-CCM+ simulation simultaneously avoiding any data mapping. In these cases, the fluid domain time step takes precedence so that the flow structures within the manifold are resolved.

Figure 1: Dr. Sinan Eroglu,

Powertrain 3D CFD Supervisor

Figure 2: Coupling strategies -

sequential coupling (left), co-

simulation (middle), conjugate

heat transfer (right)

45


Figure 3: Fluid and solid models

in STAR-CCM+ and Abaqus (top)

Figure 4: STAR-CCM+ to Abaqus

thermal data exchange

convergence (middle left)

Figure 5: Temperature results at

110 sec (middle right)

Figure 6: Warm up time for co-

simulation versus CHT method

(below)

The time step of 10 microseconds was applied to the whole model, which would have led to substantial run times for the solid model as it does not require such high temporal fidelity. Because of this, the CHT method had a 110 second time limit imposed on it.

Comparison of the methodsComparison of the three approaches over the 110 second time limit are shown in figure 5. In the neck region, the temperatures in the sequential (method 1) and co-simulation (method 2) approaches are over predicted compared to the CHT approach (method 3). Temperature time history plots at points randomly located on the manifold, shown in figure 6, demonstrate a 20°C to 25°C discrepancy between method 2 and 3 near the neck region, showing that method 2 has over predicted the temperature. However, the rate of warm up shows good agreement between the two.

Comparing to the physical test data from engine dynamometer testing in figure 7, thermal imaging results indicate that there is, at worst, a 4 percent over prediction in temperature as a result of method 2 and an average over prediction of only 1.7 percent. However, comparing method 2 thermal results to the physical data shows at worst a

9.2 percent under prediction and an average 7.3 percent under prediction. This shows that the thermal analysis coupling approach solely within STAR-CCM+ gives an overall close agreement to physical test data, albeit marginally conservative.

Piston cooling simulationGiven the close agreement of the co-simulation method within STAR-CCM+ for fluid to solid thermal analysis, this methodology was carried on to analyze the effect of the oil jet aimed beneath the piston in order to cool it down. As shown in figure 8, the oil cooling jet is aimed beneath the piston head into the piston cooling gallery so that it is closer to the critical heat source of

46

the piston crown where combustion thermal loads act on. The channel-like nature of the cooling gallery allows longer residence time of the oil which aids heat transfer. As technical specialist Serdar Güryuva points out, “As the specific power rating of diesel engines are increased by high pressure fuel injection and higher turbocharging pressures, thermal durability of the piston is increasingly important. Things such as lubricant quality deterioration - like coking, thermal cracking, carbon deposition, ring sticking and micro welding - are of concern. The lubricating oil serves as a secondary cooling medium for pistons to limit the temperature. It does so by the Cocktail Shaker Effect (CSE) inside the oil gallery, as the oil penetrates it and provides localized cooling to the piston head.”

Ground Transportation | Ford Otosan improves diesel engine design

Figure 7: Thermal Camera results

at 600 sec

Figure 8: Schematic of piston

cooling jet

The STAR-CCM+ co-simulation was carried out on a piston model that had the correct piston motion applied to it using piston velocity input data with constant temperature oil properties. An additional model was run with a static piston head and a relative motion applied to the inlet oil cooling volume so that no mesh motion was required. Only the inner liner, oil cooling jet nozzle and piston head were considered; the crankshaft and connecting rod were removed. Solid to fluid model data exchange occurred every two piston cycles, where in the last full cycle mean convective heat transfer was mapped from fluid to the solid. The thermal results from the STAR-CCM+ co-simulation were compared against physical testing data at various sensor locations.

Results for the piston with correct motion applied to it are shown in figure 9, which shows heat transfer coefficient and normalized temperature (as referenced to the crankcase temperature, T

0=Temp/

Tempcrank) for the fluid to solid interface, and temperature on the outer piston head surface. This shows the high heat transfer where the oil jet initially impinges into the oil gallery on the right hand side, and coincides with the coolest temperature within the oil gallery; T

0=1.25. As the oil

moves around the gallery, the heat transfer reduces and the temperature begins to rise again. For the complete solid piston model, the highest temperatures are inside the rim of the piston bowl, where the in-cylinder combustion thermal loads are acting. However, it is noticeable that the white contour of T

0=1.5

corresponds to the oil gallery. Correlation to the physical test results for both types of piston motion (full motion and static with oil jet relative velocity) are shown in figure 10. The results indicate very good

47


“STAR-CCM+ co-simulation enables

coupling of multi-physics simulations with

different time scales ranging from

microseconds to thousands of seconds,

providing faster and more accurate

analyses and shorter turnover times for

development and assessment of complex

designs.” - Sinan Eroglu

Figure 9: Heat transfer coefficient and normalized temperature on fluid side (left and

middle) and normalized temperature on piston solid surface (right)

Figure 10: Temperature probe point comparison between CFD and physical test (below)

agreement between the physical test and the STAR-CCM+ co-simulation.

ConclusionWith the higher thermal management demands that engines are increasingly being placed under, it is readily apparent that accurately simulating multiple solid and fluid regions can be handled quickly and accurately within STAR-CCM+. This is especially true of problems where highly non-uniform distribution of temperature and heat transfer coefficients exist, such as the exhaust manifold and piston head cooling simulations from Ford Otosan. Summarizing the results from these simulations, Sinan Eroglu said, “For the 10-minute exhaust manifold warm up case, the STAR-CCM+ to STAR-CCM+ co-simulation takes slightly longer to run than external coupling but gives more accurate results.” Due to the level of agreement in the oil jet cooling results, Serdar Güryuva is already aiming at a larger model: “As the piston wall temperatures affect the convective thermal loads, it is necessary to run a CHT analysis for the complete head and block system.”On what this means for the future of die-sel engine design, Sinan offers the follow-ing view, “The future of the diesel engine focuses on efficiency due to forthcoming CO

2 regulations, resulting in the necessity

to develop engines that withstand higher thermal and structural loads. STAR-CCM+ provides the opportunity to accurately as-sess the performance of several engine sub-systems and determine the loads and shape of the components and systems that are affected by complex multi-phys-ics.” So, through the use of co-simulation within STAR-CCM+, Ford Otosan are able to achieve better designs, faster.

48

49

Jens Dickhoff - B&B-AGEMAMasahide Kazari and Ryozo Tanaka - Kawasaki Heavy Industries

Improving cooling effectiveness of gas turbines through design exploration

Finding ways to increase temperatures at the combustor exit and high-pressure turbine stage inlet is key to boosting the efficiency of gas turbines. But higher operating temperatures jeopardize the integrity of the high-pressure turbine components, especially the vanes and blades, since modern turbine stage inlet temperatures exceed the melting points of turbine blade materials. To combat this, turbine blade designs have incorporated a technique known as film cooling.

During film cooling, cool air is bled from the compressor stage, ducted to the internal chambers of the blades and vanes, and discharged through small holes in the blade and vane walls. This air provides a thin, cool

insulating layer along the surface of the blades and vanes.

The L30A from Kawasaki Heavy Industries (KHI) is the world’s most efficient gas turbine in its 30-megawatt power class. The L30A was developed by KHI with support from B&B-AGEMA GmbH, an engineering services firm based in Aachen, Germany, specializing in the design of energy conversion machinery and plants, most notably gas turbine components. Conjugate heat transfer (CHT), a computational fluid dynamics (CFD) technique for predicting thermal flux between a solid body and a gas or liquid flowing over or inside it, is a particular expertise of the firm, which has worked closely with Siemens PLM Software

Figure 1: CFD simulation of gas

turbine blade cooling showing

a) blade cutaway view

b) cooling air pathway and

streamlines

c) blade surface temperature

Improving cooling effectiveness of gas turbines | Energy

50

to simulate 3D fluid flow and pioneer new CHT methods.

Cooperation between B&B-AGEMA and KHI began in the 1990s when KHI sought B&B-AGEMA’s help to apply CHT methods to improve internal cooling of its turbine blade designs. B&B-AGEMA developed novel film cooling technology that, instead of conventional cylindrical holes, used fan-shaped holes to direct the flow

of the air jets, thereby increasing their cooling effectiveness.

Specifically, from the 2000s on, B&B-AGEMA used CFD methods for film cooling simulations (1999-2002) and developed a technique known as double jet film cooling (1999) as well as the “Nekomimi” film cooling technology described below (2008). This work hinged on KHI’s recognition that further technological advances would require an increasing reliance on fluid thermal modeling, simulation, and design exploration.

For some years, B&B-AGEMA and KHI applied STAR-CCM+® Software to perform design space exploration manually – that is, slowly and iteratively – to study the cooling effectiveness of different shaped holes in gas turbine blades, including shapes that the two companies nicknamed nekomimi which is Japanese for cat’s ears, reflecting the visual appearance of the holes.

The computational domain used to virtually test the cooling effectiveness of different shaped holes (figure 4) consists of a main cross-flow duct and a plenum for the coolant supply, connected by the film cooling hole. The walls at the lateral sides are defined as symmetry planes in order to represent a row of film cooling holes – typical for gas turbine applications.

The plenum serves as cooling air supply for the film cooling hole. The adiabatic film cooling effectiveness has been spatially averaged on the surface highlighted in red.

Figure 2: L30A on the heavy-duty

gas turbine test rig at Kawasaki

Akashi Plant, Japan (top)

Figure 3: Film cooling hole

geometries: cylindrical (top), fan-

shaped (middle), Nekomimi

(below)

Energy | Improving cooling effectiveness of gas turbines

51

The domain width and length is equal for all configurations; this allows comparison between different cooling hole designs with similar coolant mass flow rates, as they have the same cooling air consumption per unit area.

As illustrated in figure 5, for one particular comparison, a similarly-sized Nekomimi hole results in approximately equal film cooling effectiveness at a significantly lower mass flow rate compared to the fan-shaped hole. Note that on the normalized scale from 0-1 that is typically used for cooling effectiveness, red=1 (better) while violet=0 (worse).

The result has been profound cooling improvements of 200 percent to 300 percent in the nekomimi designs over reference shaped holes – technology that has been co-patented by KHI and B&B-AGEMA.

How film cooling works and the Nekomimi hole advantageThe air used in the film cooling of gas turbines is extracted from the turbine’s high-pressure compressor, so increasing the amount of air used for cooling decreases the thermal efficiency of the turbine. Furthermore, film cooling leads to mixing losses and reduced total temperature within the hot gas passage of the turbine. These inefficiencies can be ameliorated by finding ways to reduce the amount of cooling air needed, and to establish a more homogenous solid temperature distribution.

The cooling fluid injection through a hole leads to a “jet in cross-flow” situation, shown in figure 7. Secondary flow struc-tures, including rotating vortices, are gener-ated by interaction between the coolant jet and the cross-flow which can degrade film cooling effectiveness. These degradations

Figure 4: Computational domain

used to virtually test the cooling

effectiveness of different shaped

holes (top)

Figure 5: Comparison of cooling

effectiveness and mass flow rate

for a Nekomimi versus fan-

shaped hole (below)


52

can be remedied by using a shaped-hole exit instead of a round hole, which leads to reduced momentum flux ratio between the coolant and cross flow at the cooling-hole exit (caused by the flow deceleration inside the diffusor part of the shaped hole), and a Coandă effect which facilitates the flow hugging the wall behind the hole. To reduce the undesirable mixing between the coolant and the hot gas, thus preserving a cooling layer near the surface of the turbine blade, double jet film cooling (DJFC) tech-nology was introduced by B&B-AGEMA engi-neers in 1999.

The Nekomimi technologyIn 2008, B&B-AGEMA debuted a novel hole design derived from the DJFC concept; the Nekomimi technology. This combines the

two cylindrical holes of the DJFC within a single hole design to overcome the inefficiency of the air supply situation. This was achieved by shifting the holes of the DJFC configuration to the same streamwise position (figure 9, step 1), uniting both holes (figure 9, step 2) and replacing the two supply holes with a central one (figure 9, step 3).

Automated design exploration of the Nekomimi shapeRecently B&B-AGEMA and KHI decided to automate their design search through the use of HEEDS™, the design exploration software from Siemens PLM’s Red Cedar Technology subsidiary, and the HEEDS-based Optimate+™ add-on module for STAR-CCM+. This change makes it possible

Figure 6: Film cooling

effectiveness improved more

than 200 percent from shaped

hole to best Nekomimi (top)

Figure 7: Each cooling hole is a

jet in cross-flow (below)


53

for them to evaluate hundreds of designs in the time previously required to assess just a handful, methodically comparing large numbers of traditional fan-shaped hole designs to Nekomimi-shaped holes.

Engineers from KHI and B&B-AGEMA worked with Siemens PLM to carry out an intelligent and automated search of the design landscape to identify nekomimi designs meeting conflicting objectives: low coolant mass flow rate and high adiabatic film cooling effectiveness on the test section. The parameters defining the shape of the Nekomimi holes (figure 10) were varied over 349 fluid dynamic simulations to generate a Pareto frontier of designs representing the best tradeoffs between the two objectives. Additionally, the design landscape of a laidback fan-shaped film cooling hole was searched over 299 simulations as reference in order to show the advantages of the Nekomimi technology.

Design search procedureOptimate+ was used for the automated design exploration process, STAR-CCM+ for

fluid flow and heat transfer simulation as well as geometry modeling of the fan-shaped holes, Siemens NX for parametric geometry modeling of the Nekomimi-shaped holes, and HEEDS post for visualizing and interpreting results, outlined in figure 11.

Optimate+ selects a set of design parameters and requests the CAD modeler to generate updated geometry. Then Optimate+ directs STAR-CCM+ to import the new geometry, automatically create an appropriate discretized mesh of the solution domain, and simulate the fluid flow and heat transfer. Optimate+ interactively reports the simulation results and predicted performance characteristics back to the engineer through a visualization tool named HEEDS Post.

Optimate+ intelligently uses the performance metrics to select a new set of design variables for the hole shape and repeats the process in an effort to discover better performing designs in a limited number of design evaluations. The engineer is also free to collaboratively influence the

Figure 8: Double jet film cooling

Figure 9: Nekomimi design concept:

a) step 1 (DJFC); b) step 2; c) step 3

(Nekomimi)


54

Figure 10: Nekomimi design

parameters (top); reference fan-

shaped-hole parameters (below)

“This work hinged on KHI’s recognition that further technological advances would require an increasing reliance on fluid thermal modeling, simulation and design exploration.The result has been profound cooling improvements of 200 percent to 300 percent in the Nekomimi designs over reference shaped holes – technology that has now been co-patented by KHI and B&B-AGEMA.This novel approach makes it possible to build a database of the best Nekomimi cooling-hole designs for a variety of pressure ratios and coolant mass flow rates.”


55

search by injecting designs to be evaluated based on intuition.

Design exploration resultsThe review of the results of the best possible hole shape is demonstrated by the Pareto front in figure 12 and shows the best-possible Nekomimi holes (blue dash-dotted line) and fan-shaped holes (red dash-dotted line) within the design space. These fronts show that the Nekomimi technology has significantly better spatially-averaged film cooling effectiveness for coolant mass flow rates between 8 g/s and 17 g/s. Below and above that range, both cooling hole concepts can reach comparable values for cooling effectiveness.

Also, analysis of two representative sets of simulation results (black dashed-line boxes) shows that for fan-shaped cooling holes, when the design parameters are not carefully chosen, counter-rotating vortices

dominate the secondary flow structures and worsen the cooling effectiveness. In contrast, the Nekomimi shape delivers more consistently effective cooling performance across a wide range of design parameters.

This novel approach makes it possible to build a database of the best nekomimi cooling-hole designs for a variety of pressure ratios and coolant mass flow rates. From this database, cooling-design engineers can select the best design to achieve higher cooling effectiveness and lower cooling air consumption (figures 12 and 13).

For all kinds of film cooling holes, this study strongly enhances basic understanding of secondary flow phenomena and their impact on cooling effectiveness. Further, it proves the value of automated design space exploration for solving a broad range of standard engineering problems.

Figure 11: Automated design

exploration procedure

Figure 12: Film cooling

effectiveness for all tested

Nekomimi and fan-shaped film

cooling hole designs (left)

Figure 13: Pareto front of best

Nekomimi designs as trade-off

between higher film cooling

effectiveness versus lower

coolant mass flow (right)


56

Key individuals leading the project for their respective organizations were Alain Ledoux, naval architect at Total S.A., Olivier Langeard, project engineer at DORIS Engineering, Daniel Barcarolo, PhD, senior aero & hydrodynamic and project engineer at HydrOcean, Graham Knapp, R&D engineer at CSTB, Benjamin Rousse, chief scientist at Océanide, and Olivier Bachman, support and application engineer at Siemens PLM Software.

57

Daniel Barcarolo - HydrOceanBruce Jenkins

CFD simulations of offshore oil and gas platforms are used to predict the maximum wind loads acting on the structure of the platform topsides (the upper half of the platform, above sea level and outside the splash zone, including the oil production plant, the accommodation block and any drilling equipment). The wind loads are used as waves and current loads to design the mooring of the structure, and can be significant in specific ocean areas. Because platform topside geometries were traditionally considered too complex to mesh in high detail, CFD simulations relied on simplified, de-featured models. Their low fidelity meant that, when CFD was used, the results needed to be validated and refined through expensive, labor-intensive fabrication and testing of physical models in wind tunnels.

But in a recent R&D project funded primarily by French multinational Oil & Gas company Total S.A., with engineering management and additional funding from offshore engineering firm DORIS Engineering, numerical aero & hydrodynamics specialist HydrOcean, subsidiary of Bureau Veritas Marine & Offshore, succeeded in meshing a fully detailed CAD model of a topsides module. Using this high-fidelity CFD model, HydrOcean investigated the effects of multiple wind angles on the platform, matching experimental data from wind tunnel and wave basin tests with wind generation capabilities to within 10 percent - and in most cases 3 to 5 percent - at a fraction of the cost.

The project was financed through Collaborative & Innovative Technology Program in Exploration and Production of Hydrocarbons (CITEPH), a French national program that facilitates access to private funding of innovative R&D projects in oil and gas and related energy industries (http://

CFD simulation for a fully featured offshore platform model

www.citeph.fr/en/). This was the second of two CITEPH projects in which HydrOcean received funding from Total to develop and validate the use of numerical CFD technology, specifically STAR-CCM+® Software, to compute wind loads on offshore platforms.

CITEPH Wind Loads IIn the CITEPH Wind Loads I project, completed in 2013 and funded entirely by Total, HydrOcean used CFD to reproduce results of wind tunnel experiments that had been performed for a large FPSO unit in West Africa. The project successfully validated the use of CFD for such applications, and enabled to identify the bias of experiments such as confinement, scale effects and boundary layer effects.

CITEPH Wind Loads IIWith the use of CFD validated, the CITEPH Wind Loads II project was designed to push CFD to its limits. This project, again managed by HydrOcean, required finding ways to increase the fidelity of both the physical models used in wind tunnel and wave basin testing and the numerical CFD model.Traditionally, in physical experiments used to assess wind loads over offshore structures, the detailed CAD model of the platform is simplified substantially in the process of creating a physical model for testing. Small pipes are replaced by porosity grids, while complex objects are replaced by boxes, cylinders or other simplified geometry. But in this project, the participants realized that contemporary 3D printing technology would make it feasible to produce physical models that reproduce all the details contained in the CAD model. Meanwhile, for numerical simulation, in theory, any kind of geometry can be captured with today’s CFD meshing tools.

CFD simulation for a fully featured offshore platform model | Oil and Gas

58

Hence, in this project, there were two parallel workgroups. On the experimental side were CSTB, a provider of research, testing, certification and training for built-environment and construction industries which managed wind tunnel testing for the project, and OCEANIDE, an offshore and coastal engineering firm that took charge of wind load experiments in wave basin testing.

Meanwhile, on the numerical side of the project, HydrOcean’s goal was to take a 3D CAD model of the platform topsides provided by DORIS Engineering designed for other purposes (layout, MTO) and mesh it directly without having to suppress or correct any geometry details. After months of trial and error, HydrOcean succeeded in meshing this extremely complex 3D Rhino model using the surface wrapper in STAR-CCM+, which shrink-wraps a high-quality triangulated surface onto the geometry, closing holes, joining disconnected and overlapping surfaces, as well as automatically discarding obsolete surfaces. The generated grids in this model had on average 140 million trimmed hexahedral cells.

In an intensive joint effort with the Siemens PLM Software support team in France, HydrOcean found the correct parameters that provided good mesh quality (wrapper and volume mesh) and good numerical parameters for the simulation model. The ultimate goal was to prove the feasibility of bypassing the CAD cleaning phase and use the surface wrapper instead and this was successfully demonstrated.

CFD results closely match experimental dataInitially, HydrOcean simulated seven to eight wind heading angles, and the results for the

forces were within 3 to 5 percent of experimental test data. Ultimately, the firm simulated 13 headings in all. In the largest divergence from test data, the simulation results differed by 10 percent.Next, two levels of simplification were performed on the CFD model and the corresponding 3D printed model to assess the impact of each simplification on the measured wind loads and moments. In the first simplification exercise in the 3D printed model, small pipes were replaced with porosity grids and complex objects such as pumps and valves were replaced with boxes, cylinders or other simplified shapes - typical of industry-standard practice at present. In the second, more drastic simplification, the entire platform geometry was replaced by a simple box shape. The resulting geometries were simulated numerically and also tested physically in the wind tunnel.

Both levels of simplifications compared well for the wind load forces - simulation results differed 5 to 10 percent from experimental results for the first level of simplification, and 8 to 15 percent for the second level. For the moment, however, it was harder to obtain a match, mainly for the most simplified geometry. For the most highly detailed geometry, simulation results were within 4 to 8 percent in average of test results.

No comparisons were made for the first level of simplification. For the second (most extreme) level of simplification, simulation results were within 4 to 6 percent in average of experimental results.

Could CFD model preparation be made as straightforward as converting CAD models into STL files for 3D printing?One key interest of project funders Total and

Figure 1: Views of the full-

featured CFD model

Oil and Gas | CFD simulation for a fully featured offshore platform model

59

DORIS Engineering was to learn whether preparation of CFD models from 3D CAD files could be made as straightforward as the process of converting 3D CAD files into STL files for 3D printing. From extensive prior experience using CAD data to create physical models for wind tunnel testing, the companies knew that this process - although requiring some simplification of the CAD model - is nonetheless much more straightforward than the labor-intensive geometry simplification and de-featuring traditionally required for CFD model preparation. Their goal, says HydrOcean, was to develop a comparable approach for CFD.

The 3D CAD model that HydrOcean received from DORIS Engineering in Rhino format was clean and for purposes other than CFD modeling (layout, MTO, etc). However, it presented features such as multiple disconnected surfaces and elements, and very high levels of geometric detail which made it difficult to mesh for a CFD analysis. Initially, the Siemens PLM support team suggested that HydrOcean split the model into multiple boxed regions, perform a local wrapping of each box, then perform a single wrapping operation to merge all the boxes. But HydrOcean decided that approach would require an unrealistic amount of time and labor and would not be the industrially feasible method sought by Total and DORIS.Instead, HydrOcean decided to invest considerable time and research in finding how to apply the surface wrapper tool to accomplish meshing of the entire geometry in a single process, without splitting the geometry and significant loss of detail. The majority of time was spent on finding how to create a wrapped surface that could generate a very good volume mesh.

The quality of the simulation solution, as confirmed by comparison with experimental

data, will be the confirmation that HydrOcean had found a very good wrapping setting and methodology. Initially, the solution did not converge. Upon closer examination, HydrOcean found the simulation model showed some areas of very high pressure, high velocity profiles and other results that revealed some regions of the volume mesh were not of high quality. Those results, it says, were caused by the wrapper, so it explored how to adjust the wrapping process to create a volume mesh that would yield a simulation that converged. Over some 15 iterations, numerous surface parameters were varied, seeking to find the best connection among elements while avoiding such things as suboptimal triangulation areas in the mesh. In the end, it found surface parameter settings that yielded simulation results that in all instances were within 10 percent of experimental values.

Thus, HydrOcean delivered an industrially feasible CFD simulation approach for full-scale, high fidelity, fully featured offshore platform modeling with STAR-CCM+, which will lower prototype and testing costs in the industry.

Figure 2: Flow simulation results using full-featured CFD model Figure 3: Full-featured 3D printed model used in wind tunnel and

wave-basin testing

Figure 4: Correlation of CFD

simulation versus physical test

results (dots show simulation

results, dashed line shows test

results)

CFD simulation for a fully featured offshore platform model | Oil and Gas

60

61

Alex Read - Siemens PLM Software

CFD for separatorsIf you work in the field of process and separation, chances are you have come across computational fluid dynamics (CFD). CFD produces a wide range of emotions ranging from abject fear, often involving flashbacks to a long forgotten university class and a dizzying array of partial differential equations, to curiosity and even enthusiasm. The purpose of this article is to allay those fears, answer some questions and help you become an educated consumer of CFD.

Many would start by asking the “so what” question: Why and when should CFD be cared about? This question will be answered, followed by a brief introduction to CFD including the major multiphase models, answering some frequently asked questions (FAQs) and ending with a short case study example of CFD and automated design exploration being applied to a cyclone separator, all of which will be achieved without recourse to a single equation!

So why should CFD be part of the separator design?

In today’s “lower for longer” market, cost reduction is front and center for all. CFD helps in all the phases of a project: from

Savvy separatorsIntroduction to computational fluid dynamics for separator design

reducing the initial project costs (CAPEX) and operating costs (OPEX) to helping manage project extensions, such as tying in additional wells to an existing facility.

A subsea separator’s job is to separate gases and liquid prior to pumping. Ensuring the separator meets its process requirements is important. The pump or compressor downstream will not perform as desired if there is carry-over or under (liquid in the gas stream and gas in the liquid stream). Should this occur, operations will have a very expensive problem to resolve.

In addition to meeting the minimum process requirements, there are other design considerations. There may be a need to reduce its weight to ease installation, understand how changes in upstream piping impact performance to provide a standardized design able to connect with many Subsea Processing Systems (SPS) configurations, minimize its size to reduce the amount of real estate used, minimize the pressure drop and the use and cost of internals.

Hand calculations, such as Stokes law, can be used to estimate the required residence time, but this involves assumptions about the flow (for example, no short-circuiting

Savvy separators | Oil and Gas

62

or even distribution across the flow area). Physical tests can be run, however these present their own challenges: cost and time, difficulty visualizing the complex multiphase flow, testing with process fluids and high pressures and temperatures introducing significant additional cost and safety concerns.

Using CFD, the flow patterns in upstream piping, inlet devices and within the separator can be predicted to ensure adequate residence time. Each separation mechanism can be studied such as estimating the tendency for liquid droplet re-entrainment due to high gas velocities and improving the performance of internals by increasing the uniformity of the flow to the demisters (see Dr. Lee Rhyne’s SPE webinar “CFD Optimization of Scrubber Inlet Design” for examples).

CFD achieves this with relative ease and speed, using actual process fluids, temperatures and pressures. After simulating the initial design, we can run “what-if” design studies to see how the design can be improved. In short, successful use of CFD results in discovering better designs, faster and at a lower cost.

This is predicated on the assumption that the CFD study was conducted correctly, and therefore the results can be trusted. Many of the uncertainties of CFD can be systematically tested for (such as a mesh refinement study), or can be interrogated by an engineer who understands the relevant physics at hand (and not CFD). A common concern is that results must be “tuned” to be accurate. This is not necessarily the case as accuracy can be obtained through systematic testing, quantification and minimization of uncertainty and error, and by ensuring the problem is adequately posed and modeling

assumptions are appropriate for the problem at hand. The first step in interrogating a CFD solution is to use good engineering judgment. Does the flow field make sense, and if not, why? How does it compare with hand calculations, prior designs that have validation data and simpler analysis methods? In the early stages, CFD mistakes are often typos, so if it looks like the code solved a different problem to the one being investigated, there’s a good chance it did!

Next, it takes a deeper explanation to understand the major steps in the simulation process and the potential impact of these on the results.

CFD attempts to solve the Navier-Stokes equations which describe the behavior of fluids. Unfortunately, solving the Navier-Stokes equations is computationally intractable, so for all practical problems, the Reynolds Averaged Navier-Stokes (RANS) form is used. These suffer from the “closure problem” as they have more unknowns due to averaging which are resolved through the use of turbulence models. Models are also used to incorporate more advanced physics, such as multiphase flow dynamics.

Setting up a CFD study involves four steps: 1. Defining the domain of interest,

geometry, flow entry and exit and boundary conditions (for example, velocity at the inlet)

2. Discretizing or meshing the domain: Rather than solving for the fluid behavior (velocity, pressure etc.) at every point in space, we segment the volume using a mesh, then solve the equations only at the center of each cell in the mesh - think of a separator filled with Lego bricks. CFD will tell you what the velocity, pressure, temperature and so on, is at the center of

Figure 1: Different cell types:

tetrahedral (blue), hexahedral

(green), polyhedral (red)

Oil and Gas | Savvy separators

63

each brick. For completeness, in time-varying problems, time is discretized by solving for time increments, such as every 0.1 seconds.

3. Selecting the physics to simulate, whether the flow is single phase or multiphase, thermal or isothermal.

4. The CFD program then iteratively solves the equations to convergence.

These steps also represent the four areas of uncertainty and where attention should be focused.

If a 3D CAD model or drawings of the geometry exist and are available, defining the domain typically is not an issue. Defining the flow conditions at the inlet and outlet is more challenging. In separator simulations it is typical to prescribe a droplet size distribution at the inlet, which requires knowing or estimating this. Typically, boundary condition information is available from other analysis methods – for example, from a 1D model of the system, from CFD by extending the domain of interest upstream to some point where there is less uncertainty or by hand calculations, for example, to estimate minimum droplet sizes. The engineer who performed the CFD study should be able to explain what conditions were used, what these mean physically and why they are appropriate.

Next, the domain (geometry) needs to be discretized or meshed. This is an important and potentially time-consuming part of the analysis. A good mesh begets good CFD.

The mesh (number, size, and type of cells used) can influence the answer. As an example, if a CFD simulation is trying to simulate a vortex using one cell, the solver only has one point at which it calculates the velocity and pressure to represent the

vortex. As the number of cells is increased, decreasing the size of each cell, the resolution and therefore accuracy of the representation of the vortex improves. A similar approach is used to ensure or minimize the influence of the mesh over the solution, the mesh is progressively refined (reducing cell size) until quantities of interest, such as pressure drop, stop changing. A short note on a well-established method for grid convergence studies can be found at

1.

There are also different cell types: hexahedral (six sides), polyhedral (many sided, but typically soccer ball shaped with 12-14 sides) or tetrahedral (four sides).

Historically, tetrahedral meshes were often used, since building hexahedral meshes was difficult and time consuming, particularly for complex geometries. This is undesirable as tetrahedral cells have poor numerical properties, they artificially make the fluid behave as if it is more viscous. As a consequence, many more tetrahedral cells are required to attain the same level of accuracy as when using hexahedral or polyhedral meshes.

Fortunately, meshing packages have improved significantly in the last decade, so it is now possible to build polyhedral or hexahedral meshes, even on complex geometries, without significant overhead.

After building the mesh, the engineer must select the physics to consider. One benefit of CFD is the ability to simplify problems to consider only the physics of interest, making it easier to interrogate and understand results and trends. This is also a double-edged sword as over simplifying can miss important effects. As with boundary condition selection, the engineer performing the analysis should be able to

Figure 2: Droplets modeled using

VOF (left)

Figure 3: Geometry and flow

visualization for baseline design

(right)


64

explain the models used, their physical meaning and appropriateness for the problem at hand.

In the separator world multiphase modeling is key. There are three main multiphase models used in CFD:1. Free-surface or Volume Of Fluid (VOF)2. Eulerian-Lagrangian Multiphase often

shortened to Lagrangian Multiphase or LMP

3. Eulerian Multiphase (EMP)

In the VOF approach, the interface between the phases is resolved with the mesh. As in figure 2, if droplets of water fall under gravity through air, CFD can capture the motion of the droplet using the VOF model if there is adequate mesh resolution to capture the shape and motion of the droplet in the mesh.

Consequently, this model is well suited for flows with a well-defined interface between the phases, such as stratified flow, where the mesh can be refined locally to capture the interface. A common application of VOF is to model the sea and its behavior around ships - there is a clear interface between the sea and the air.

VOF can be used to model flows other than stratified but the mesh needs to be refined to capture the multiphase effects at the interface between the phases, such as entrainment of fine droplets into the gas phase. However, this increases the computational cost of the analysis. In the separation world, VOF is often

used to evaluate the bulk flow properties of the vessel.

When the flow is dispersed, either Lagrangian Multiphase (LMP) or Eulerian Multiphase (EMP) is typically used.

In LMP, the continuous flow field is solved using the RANS CFD approach. In the example of droplets falling through air, the air is the continuous phase and the droplets are the dispersed phase. The continuous phase is solved in an Eulerian framework with a fixed mesh and the flow motion relative to the mesh. The full name for LMP is Eulerian-Lagrangian multiphase, but the Eulerian is dropped for expediency.

For the dispersed phase, the trajectory of each particle or droplet is solved for using Newton’s second law of motion. The calculation of the droplet or particle motion is performed from the reference frame of the moving droplet, rather than the fixed mesh, which is known as a Lagrangian method. In order to reduce the computational cost and make it applicable to scenarios with a large number of droplets, each droplet represents an ensemble of droplets. Inputs to the motion calculation include sub-models for the drag force and dispersion of droplets or particles due to turbulence. Additional sub-models can be introduced to include breakup and coalescence of droplets. The interaction between the phases can be either one- or two-way. One-way is where the motion of the droplets is influenced by the

Figure 4: Baseline cyclone

geometry


65

continuous phase but the continuous phase does not “see” the droplets, two-way is where both phases influence each other. The one-way coupling is often applied.

This method is an efficient and accurate way to model droplet or particle flow but is less applicable when the volume fraction of droplets or particles is high. Opinions on the volume fraction cut-off vary but are usually in the 5-10 percent range, at which point the model accuracy and stability deteriorate. For particle flows, more advanced models can be used to address this, such as Discrete Element Method (DEM) or Multiphase Particle-in-cell (MP-PIC). In the separator world, LMP is often used to study the motion of droplets in the gas stream.

For Eulerian multiphase, the full RANS equations are solved for each phase. Using the concept of “interpenetrating continua,” the continuous and dispersed phases interact through source terms for drag, lift, virtual mass and turbulent dispersion. This makes it an immensely flexible model, able

to simultaneously handle any number of phases and any range of volume fractions. Sub-models can be included to account for additional physics such as breakup and coalescence of droplets or bubbles, or heat and mass transfer.

The disadvantages of EMP are that each set of RANS equations comes at a cost (studying many particle sizes or phases becomes computationally expensive) and the user needs to understand, and choose, appropriate sub-models and settings. The downside of EMP’s flexibility is that it can be applied to a wide range of multiphase flows, which results in multiple sub-models to be understood and applied.

For analysis of separators, EMP can be used to model the full vessel, but is particularly effective in mixing regions where volume fractions exceed the limitations of LMP and in tracking small droplets or bubbles with VOF is computationally expensive.

Having built the mesh, specified boundary conditions and chosen appropriate

Figure 5: Geometry of improved

design, velocity magnitude along

a plane section through the

center of the cyclone and design

study history showing the

progressive improvement in

separation efficiency


66

modeling assumptions, the solver uses iterative techniques to successively improve the solution until “convergence” is attained . Mathematically, convergence describes the limiting behavior, particularly of a series towards its limit. In CFD, the series is the flow field (values for velocity, pressures etc.). The flow field reaches its limit when the values for velocity, pressure and so on, stop changing from iteration to iteration.

Convergence is often judged by monitoring residuals. Residuals measure the amount by which the discretized equations are not satisfied. A typical rule of thumb is that residual values should have dropped by three orders of magnitude.

If the residual values do not drop and the flow field continues to change, this may indicate that the steady-state assumption does not work due to inherent unstable phenomena like turbulence. Alternatively, it may indicate problems such as poor quality cells in the mesh or an ill-posed problem such as the location or values of boundary conditions.

Case study: Design space exploration of a gas-solid cyclone separator to improve separation efficiency Having described the steps in setting up a CFD study and building confidence in the result, the following is an example of how

CFD should be used to understand and improve the design and performance of a gas-solid cyclone separator.

The CFD simulations were run using STAR-CCM+®, a Siemens PLM software. Fi-gure 3 shows the geometry of the separa-tor, streamlines through the device with an isosurface showing areas of low pressure, volume rendering of pressure contours and a comparison between CFD using multiple methods and experimental data for mean axial velocity at two locations in the cyclo-ne. The purpose of these simulations was model verification, which is why multiple methods were used for the same case.

The baseline geometry for the design study is from an European Research Community On Flow, Turbulence and Combustion (ERCOFTAC) paper where the geometry and experimental data - Laser Doppler Velocimetry (LDV) - of the mean velocity profiles across the cyclone are available. Having validated the base model, CFD allows us to explore design alternatives quickly and easily. It can also be used in conjunction with tools that will automate the simulation process and efficiently explore the design space. In this case, the HEEDS multidisciplinary design exploration software from Siemens PLM, in conjunction with STAR-CCM+, was used.

For the design space exploration study, a constant gas velocity of 25 m/s was

Figure 6: Correlation plot


67

applied at the inlet. Sand particles with a diameter of 1.2 µm were introduced at the inlet, such that they accounted for one percent of the volume fraction of the flow. LMP with a one-way coupling to the continuous phase was used, along with the Reynolds Stress Turbulence model (RSM). The automation and design space exploration tool HEEDS™ uses algorithms to predict the next exploration point in the design space. HEEDS requires the engineer to provide:• Design objective(s): There can be more

than one and these can be competing. In this case, the objective is to maximize the separation efficiency of the cyclone.

• Constraints to be applied: In this case, the pressure drop across the separator cannot exceed a certain value or else the design will be deemed infeasible.

• Different load cases to be evaluated, for example if the separation performance is different at different particle loadings.

• Design variables: Their extents and sensible increments are to be defined. In this case, we vary the radius of the cyclone, the length of the constriction and parallel wall sections.

Many different approaches have been developed to help explore the design space efficiently. These are often referred to as optimization algorithms and include among others Design of Experiments (DOE), genetic algorithm, downhill simplex and particle swarm. Optimization is often a misnomer since for most industrial applications no single optimal solution or design exists. However, these techniques can be highly effective in identifying better designs.

An impediment to the application of these methods is their multitude: Users must understand which method to use for any given scenario. HEEDS uses a hybrid and adaptive algorithm called SHERPA, which will switch between different exploration methods (DOE, genetic algorithm and so on) depending on the information it has about the analysis such as the number of variables and time and resources available. The benefit of using these methods is that they find better designs in less iteration

than an engineer on their own or other optimization methods. In this case, 125 designs were evaluated over five days (always running two cases concurrently). After 44 tries, the separation efficiency improved by 19 percent, while increasing the pressure drop by ~1 percent.

The correlation plot shows the relationship and degree of correlation between two variables, such as the radius of the cyclone and the cyclone separation efficiency. The numbers on the top right-hand side show the correlation between the two variables represented in the square (1.0 indicates perfect correlation).

The correlation plot helps the engineer to interrogate large amounts of data (100-plus designs), and to understand quickly what influences the design. In this case, the cyclone radius has a significant impact on its efficiency and pressure drop (correlation of 0.74 and 0.71 respectively).

SummaryTo the non-specialist engineer, CFD can be initially daunting, particularly in more advanced areas such as multiphase flow and separation. While detailed knowledge of sub-models will remain with the specialist, non-specialist engineers can critique CFD results by evaluating whether the physical meaning of the results be explained and asking about the modeling decisions taken and their anticipated influence on the results and quantities of interest. By introducing the main multiphase models used, this article aims to help in this process.

CFD compliments other analysis methods (analytical or experimental). Its successful application can have a significant, positive financial impact on projects: by reducing the cost of design, improving and validating equipment performance and mitigating problems before they occur. Linking CFD with automated design space exploration tools can further the understanding and improvement of separator designs.

References1 http://journaltool.asme.org/templates/jfenumaccuracy.pdf2 https://www.nafems.org/join/resources/cfdconvergence/Page0/3 http://www.ercoftac.org/


68

Figure 1: KivuWatt methane extraction platform on Lake Kivu

69

Degassing Africa’s Lake Kivu | Oil and Gas

Nicolas Ponchaut, Timothy Morse and Gary Bigham - ExponentJavier Castro - Siemens PLM SoftwareBruce Jenkins

IntroductionLake Kivu is an African rift valley lake lying between Rwanda and the Democratic Republic of Congo (DRC). One of several East African rift lakes that hold high concentrations of dissolved carbon dioxide, Lake Kivu is unique in containing methane as well. The carbon dioxide comes from subaquatic springs that feed it into Lake Kivu’s deepest waters. The methane comes from bacteria decomposing organic matter at the lake bottom, from the springs, and from bacteria converting carbon dioxide into methane. The gases are kept in solution by the hydrostatic pressure in the deep waters.This poses a danger because a landslide or volcanic activity, or even the ever-growing concentration of dissolved gases, could cause a vertical disruption of the lake water column. In such an event, also known as a limnic eruption, gas-laden deep water is displaced to shallower depths and thus lower pressures, allowing the gas to bubble out of solution and trigger a gas eruption – an event that would imperil the lives of the more than 2 million people living along the lake’s shores.Such an eruption happened in Cameroon twice in the 1980s. The deadlier of the two came in 1986, when a cloud of carbon dioxide erupted out of Cameroon’s Lake Nyos in a 100-meter column of water, asphyxiating more than 1,700 people as far as 25 kilometers from shore. Lake Kivu, holding one thousand times more gas than Nyos, presents an even greater danger.But the government of Rwanda saw Lake Kivu’s methane-laden waters as not just a risk to be mitigated, but a resource to be exploited. It approached the U.S. energy

Degassing Africa’s Lake Kivu for public safety and power generation

services firm ContourGlobal to design and construct a system to extract the methane and use it to greatly expand the country’s electric generating capacity, while at the same time stabilizing the lake against future overturns. The result is a 300-ton floating facility that extracts gas-laden water from deep in the lake, separates the methane and some of the carbon dioxide, then injects the degassed water back into the lake.The $200 million KivuWatt biogas project, consisting of the gas extraction platform and

Figure 2: Lake Kivu, Rwanda, and

the location of the barge for the

extraction facility

70

an initial 25 megawatt generating plant, began operating on December 31, 2015. At its formal inauguration in May 2016, Rwandan president Paul Kagame announced the project will scale up to 100 megawatts by 2020, nearly doubling the country’s total generating capacity.This will be a boon to Rwanda’s economic development. Today its capital city, Kigali, is the nexus of a dynamic national economy whose yearly growth has averaged a robust eight percent over the past decade. But as the country and its capital have developed, its electricity supply has failed to keep pace. Total installed capacity before KivuWatt was only 156 megawatts. Close to 80 percent of its 12 million citizens still have no connection to the grid and those who do have faced high electric costs because of the country’s dependence on imported diesel and heavy fuel oil to power its generators.

CFD used to analyze degassed water plume and assess risk of lake overturnAs part of the KivuWatt project, engineering and scientific consulting firm Exponent, Inc. applied computational fluid dynamics (CFD) to analyze the degassed water discharge plume from the gas extraction facility and to characterize the dynamics of the plume. Two key concerns were raised with the degassed water plume: Could there be recirculation between the water intake risers and the degassed water discharge points? And could the dynamics of the plume lead to an overturn of the lake and a catastrophic gas eruption? Exponent’s simulations showed

that the degassed water plume ultimately stratifies within a density gradient, that recirculation does not occur and that the discharge plume does not result in uplift or overturn of the lake for the conditions evaluated.

Characteristics of Lake KivuLake Kivu has an unusual thermal structure. At depths below approximately 80 meters, the water temperature increases with depth. However, the water also contains large quantities of dissolved carbon dioxide (CO₂) and salt, with concentrations that increase with depth. The density increasing effects of CO₂ and salt maintain the lake stratification despite the inverted temperature profile. The density structure of the lake is a series of relatively homogeneous mixed zones separated by density gradient layers.The gases in Lake Kivu are kept in solution by hydrostatic pressure at depth. If a disruption to the lake due to a landslide or volcanic activity caused a parcel of gas-laden water to be brought to shallower depths where the hydrostatic pressure is below the saturation pressure (the pressure necessary to keep all the gases in solution), gas bubbles would form in the water. The bubbles would form a buoyant plume that would carry even more gas-laden water to shallower depths and release even more gases. This could eventually result in a large-scale release of gases at the surface of the lake.The highest concentrations of methane (CH₄) in Lake Kivu are at the lowest depths, in the upper resource zone and the lower

Figure 3: Vertical density profile

in Lake Kivu, with the different

zones labeled

Figure 4: Diagram of the floating

gas extraction barge (not to

scale)

Oil and Gas | Degassing Africa’s Lake Kivu

71

“But the government of Rwanda saw Lake Kivu’s methane-laden waters as not just a risk to be mitigated, but a resource to be exploited.”

Figure 5: Computational domain

for the plume simulation

Figure 6: Passive scalar values for

the degassed water plume: The

four discharge risers are visible

with the location of the intake

points also shown. The simula-

tion vertical symmetry plane is

through the center of the image.

The dark blue line shows the 95

percent dilution contour

resource zone. The gas extraction facility, contained on a floating barge, extracts gas-laden water from the lower resource zone (depth of 350 to 360 meters) through four extraction risers. The facility then separates the dissolved gases (CH₄, CO₂ and some H₂S) from the water and re-injects the degassed water back into the lake through four discharge risers.

Hydrodynamic simulationsTo answer the key questions – will the degassed water recirculate back into the intake risers and prevent effective gas extraction, and will the degassed water plume upset the lake stratification and lead to a gas release – Exponent conducted several simulations of the degassed water plume using STAR-CCM+® software. The density of the water, and hence the plume dynamics, is heavily affected by the water temperature and by the concentration of methane, carbon dioxide and salt. Thus the computational method tracked these quantities throughout the simulation. Background conditions in the lake, including mixed zones separated by density gradients, were also included in the model. A Reynolds-Averaged Navier Stokes (RANS) model was used, with a k-omega turbulence model. The computational domain was a cylinder with radius of 120 meters centered on the risers. The top of the domain was at a depth of 240 meters, the bottom of the domain was the lake bed at 365 meters.Pressure boundaries were used on the sides and top of the domain to allow water flow in

and out of the domain. A vertical symmetry plane was applied between the two sets of intake and discharge risers. Water temperature, pressure and species mass fractions (water, salt, CO₂ and CH₄) were imposed as the initial state, as well as inflow conditions on the pressure boundaries. The computational domain was discretized using a polyhedral mesh that was refined around the risers and diffusers. The simulation assumed no horizontal currents.

Plume densityThe density of the plume, both at the point of discharge and as it evolves and dilutes in the lake water, is the most important parameter that dictates the plume behavior. The density depends primarily on four variables: water temperature, salinity, concentration of CH₄ and concentration of CO₂. Higher temperature decreases density and while higher salinity increases density.The gas concentrations have a complicated effect on density. If the gas concentration is small enough that the local hydrostatic pressure is sufficient to hold the gases in solution, then the dissolved gases have a relatively modest (but still important) effect on the density. The presence of dissolved CO₂ increases the density, while the presence of dissolved CH₄ actually decreases the density.However, if the gas concentration is high enough that the gases cannot be kept in solution by hydrostatic pressure, bubbles will form in the water, dramatically decreasing local density. Both effects – dissolved gases and gases in bubble form – must be

Degassing Africa’s Lake Kivu | Oil and Gas

72

accounted for to correctly calculate the plume density.

Simulated plume behaviorIn the extraction scenario that Exponent examined, raw water was extracted from a depth of 355 meters and degassed water was reinjected at 280 meters. The four discharge risers terminated with a diffuser that directed the discharge flow horizontally. Simulation results are shown in figures 6 to 8. To illustrate the characteristics of the discharge plume, the concentration of a passive scalar tracer (set to 1 for the degassed water discharge and 0 for the ambient lake water) is shown in figure 6.

ConclusionIn addition to answering the critical questions, Exponent found that the

Figure 7: Density of the degassed

water discharge plume: The dot-

ted line indicates the simulation

vertical symmetry plane

Figure 8: Streamlines of the de-

gassed water plume: The view is

rotated 90 degrees from the view

in figures 6 and 7

Dr. Ponchaut is a Managing Engineer in Exponent’s Thermal Sciences practice. He specializes in the area of computational fluid dynamics (CFD) and heat transfer modeling. He has performed numerous analyses for the oil and gas industry, such as characterizing the dynamic and acoustic behavior of various pipe systems, and simulating the consequences of LNG spills, flammable vapor clouds, underwater hydrocarbon plumes, and the effect of ground flares on the surrounding temperatures and pollutant concentrations. He also has significant experience in simulating the behavior of thermal management systems to quantify the risk and propagation of thermal runaway failures in lithium-ion battery packs.

Dr. Morse is a Managing Engineer in Exponent’s Thermal Sciences practice. He specializes in the engineer-ing analysis and experimental testing of thermal and flow processes and equipment. His project experience has included turbines, compressors valves, heat exchangers, boilers, cryogenic liquids and medical devices among others. He also has expertise in the investigation and prevention of fires, explosions, and equip-ment failures and has conducted numerous fire origin and cause investigations. Dr. Morse has investigated flow-induced vibration issues over a wide range of applications including heat exchanger tube bundles, pipelines and offshore structures.

Mr. Bigham is a Principal Scientist in Exponent’s Environmental & Earth Sciences practice. He specializes in the evaluation of transport, fate and effects of contaminants in the environment. He has managed and been the principal investigator of field, laboratory, and modeling and non-modeling assessments of a wide variety of contaminants in lakes, rivers, estuarine waters, ocean waters, groundwater and air.

degassed water plume does not enter the intake risers but rather stratifies above them; and the dynamics of the degassed water plume do not cause uplift of the water column, so the gas extraction process does not present a risk of lake overturn, for the extraction scenario that was modeled. The simulation results also showed that the degassed water stratifies within a density gradient. This is because if the degassed water is discharged outside of a density gradient (where there is almost zero change in density with depth) and is denser than the surrounding water, there is nothing to stop the plume from sinking until it reaches a density gradient, regardless of how much mixing occurs. By adjusting the discharge depth, the degassed water plume can be controlled to stratify within any density gradient of choice.

Oil and Gas | Degassing Africa’s Lake Kivu

73

STAR-CCM+: Discover better designs, faster.Explore designs digitally to innovate while reducing cost and overdesign.

siemens.com/mdx

74


75

Multi-scale modeling of furnaces and reformers | CPI

Ravindra Aglave, Niveditha Krishnamoorthy, Megan Karalus, Karin Frojd and Thomas Eppinger – Siemens PLM Software

Figure 1: Various design

considerations in a reformer

The design of furnaces, process heaters, crackers, and reformers is very challenging. This is due to the high temperature environment and complex chemical reactions occurring at various spatial and temporal scales. Optimum design and operation needs to balance thermal, environmental and process performance simultaneously. Furnace and cracker operations involve the processing of large volumes of feedstock, and even small

Multi-scale modeling of furnaces and reformers

improvements in equipment and process efficiency can translate into improved operational efficiency and significant cost savings. Computational fluid dynamics (CFD) can help in the design and operation of plants and processes to achieve these goals.

Heaters and reformers have two major components: the firebox where combustion is occurring and the process side tubes where the feed is being converted to

76

CPI | Multi-scale modeling of furnaces and reformers

products. The design challenge arises due to the functioning of multiple components: multiple burners in the firebox side, multiple tubes in the process side, catalytic reactions, and thermal processes inside the coils. The performance considerations in CFD-based design include efficiency, emission control, and uniform heat distribution and conversion rate in the processes.

In this article, we explore the application of CFD to various components of furnace and reformer design using a hierarchical approach. These hierarchical scales of processes (a combination of geometrical and physics scales) are shown in figure 1. Overall, one can classify them as:

Firebox modelingThe firebox houses the burner and requires uniform fluid flow and heat distribution to

achieve good operating efficiency. Figure 2 shows a typical burner with geometric scales ranging from 0.0016m to 25m, a factor of 15,000. The first modeling challenge in such a geometry with wide ranging scales is to create a high quality mesh that takes a short amount of time to generate. A good mesh should have refinement to capture the small geometric details but be coarse enough to keep computational times reasonable. The meshing capability of STAR-CCM+® Software (polyhedral or trim depending on the need) and automated prism layer meshing ensures that the burner geometry can be easily meshed. The automatic meshing capability which allows customization of refinement regions ensures such a mesh can be created with little manual input. Qualitative and quantitative burner performance characteristics can be assessed from the results, allowing the designer to analyze oxygen concentration, temperature uniformity, recirculation zones, and flue gas engulfment. Further, the radiative and convective heat transfer duty of the tubes can be evaluated as well. Poor temperature uniformity can lead to high temperatures on the tubes, in turn leading to excessive coking and poor performance of the heater, requiring an earlier than scheduled downtime. A full suite of combustion models, multi-component species modeling and radiation modeling allow for all the physics to be accounted for. Figure 3 shows

Figure 2: Burner range of

geometric scales and mesh (from

0.005m to 15m) (top)

Figure 3 : Temperature

distribution and radiative

absorption in the burner (below)

Macro Scale Models Meso Scale Models Micro Scale Models

Firebox combustion modeling for thermal performance

Catalyst shape and packed bed modeling for yield

Reaction kinetics and reaction mechanism tuning

Burner design and optimization for flame shape and pollutants

Coupling to reaction kinetics on process side for thermal performance

Air supply duct optimiza-tion of uniformity of airflow

77


sample results from STAR-CCM+ showing the temperature distribution and radiative absorption in the burner and heating tubes.

Heater duct optimizationThe next step is to design the heater ducting with minimal mass flow variation through the burner throats and minimal pressure drop. A properly optimized heater duct design (using STAR-CCM+) shortens the product development cycle and reduces the risk of poor performance. The heater consists of a central duct connected to the burners via short legs. For this example, the radius of the connector, width and height of ducts were chosen as the design parameters (figure 4) for optimization analysis.

The heater performance was simulated in STAR-CCM+ and design optimization was conducted with HEEDSTM software, driven by the hybrid adaptive SHERPA algorithm. HEEDS drove the parametric CAD geometry and STAR-CCM+, performing 148 design evaluations by varying the design parameters. The baseline design analysis

took 40 minutes on eight cores and the entire design optimization was performed in 32 hours on 40 cores. Siemens PLM’s Power Token licensing provided complete flexibility to use the most efficient combination of parallel evaluations and solver cores for the problem at hand. Mass flow and pressure changes were used as benchmarks for the best design. Figure 5 shows the results from HEEDS evaluations, showing the design with the optimum combination of mass flow change and pressure change within the heater. Thus, STAR-CCM+ and HEEDS optimized the heater duct design within a few days, saving valuable time and cost and reducing the number of prototypes and testing while improving performance.

Process burner optimizationBurner design can affect the flame shape and length, which in turn changes the temperature distribution in the furnace. The resulting heat density effect alters the flue gas recirculation patterns. In low NOx burners, this will also change the NOx emission levels. As such, an optimized

Figure 4: Design parameters for

heater duct optimization analysis

(top left)

Figure 5: HEEDS results showing

best combination of mass flow

and pressure drop in the heater

(top right)

Figure 6: Process burner design

parameters for optimization in

STAR-CCM+ (below)

78

burner design is critical to maintain peak furnace performance from both thermal and environmental perspectives. Here, a full design optimization of a single flat flame burner using STAR-CCM+ and Optimate+, a plugin that utilizes the SHERPA search algorithm of HEEDS, is described. The design objective is to reduce the CO and NOx relative to the baseline design. The dependence of the flame height and emissions on burner design parameters like tilt angle and fuel port spacing is explored. A fully optimized burner design should have good flame stability, an acceptable flame pattern, length/shape, no flame impingement or flame to flame interaction, good combustion efficiency, and meet stringent emission standards.

The burner has four fuel ports with fuel and oxidizer entering at standard temperature and pressure. The burner heat release is approximately 0.9MW. Figure 6 shows the six design parameters used in the study. The simulations were run with the K-Omega SST turbulence model, the Presumed Probability Density Function (PPDF) equilibrium combustion model, NOx PPDF flamelet and the Discrete Ordinates Radiation Model.

Design constraints were built into the model to not exceed a volume average temperature greater than 1350⁰K in the firebox and a flame height larger than approximately 1.7m based on process heating needs and design of the furnace. Optimate+ automatically varies the design parameters and a total of 85 design evaluations were run; 13 designs did not meet the constraints and were rejected. The best performing design (figure 7) minimized the CO by 4.5 percent and NOx by 13 percent compared to the baseline results. This design had a tilt angle of 16 degrees compared to a tilt angle of 1 degree for the baseline case. The design also showed a 3 percent flame height reduction. A design with a tilt angle of 24 degrees (Design 32) showed lesser CO and NOx variation but with similar flame height. Table 1 shows a comparison of the baseline case with the best designs from Optimate+.The influence of fuel injector spacing was also investigated in the optimization analysis. The baseline design had a spacing of 88mm and the best performing designs had a spacing of 25mm (Design 17) and 250mm (Design 28) respectively. Figure 8 shows the comparison of flame

Figure 7: Pareto front and

comparison of flame height for

baseline and best designs for tilt

angle changes (left)

Figure 8: Flame comparison

between baseline and best

designs for different injector

spacing (right)

Design 1 Design 13 Design 32

Flame Height (m) 2.423 2.342 2.422

Flame Volume (m3) 2.15E-01 1.86E-01 1.84E-01


Table 1: Comparison of baseline

and best designs

79

height and volume for these designs. A 5 percent reduction in CO and 3.4 percent reduction in NOx was achieved for Design 28 but the flame height reduced by 28 percent.

The design optimization study from changes in tilt angle and injector spacing showed that Design 32 was the best design overall, showing a 4 percent CO reduction and 11 percent NOx reduction from the baseline cases.

Reacting channel co-simulationThe next component in the furnace design optimization is the process tubes in the reformer. Full 3D modeling of the process side with tubes is computationally expensive since there are multiple burners (firebox side) and tubes (process side). A computationally efficient method is co-simulation in STAR-CCM+, with 3D modeling of the firebox side coupled with 1D modeling of the process side.

The firebox side is modeled in 3D with turbulent flow, fuel/oxidizer boundary conditions, combustion models and participating media radiation model, accounting for full heat transfer. On the process side, any number of tubes can be accommodated and the tube is modeled as a 1D plug flow reactor (PFR), along with complex tube side reactions and heat exchange between firebox and process side. No meshing is needed for the 1D model saving valuable computational time. For the coupling, temperature is provided to the process side from 3D simulations and heat flux is provided to the firebox side from 1D analysis.

Figure 9 shows a sample output from co-simulation showing temperatures on the firebox side. For kinetics of process side reactions, detailed, reduced, or user-defined mechanisms can be used.

Packed bed heat transfer modelingThe design of a packed bed reactor should consider the influence of packing geometry on fluid flow, heat transfer, and pressure drop. This requires a reasonable amount of experimental trial and error to investigate kinetics, heat transfer, and fluid flow behavior to confirm a final reliable design. With the Discrete Element Modeling (DEM) capability in STAR-CCM+, industrial scale simulation of packed beds is now possible by fully resolving the geometry of the beds, enabling reliable parametric analysis for reactor design and providing detailed insight into physics like local velocity and recirculation.

Generating a representative bed could be done manually, with the Monte-Carlo method or with a scan of the bed. All these methods are time consuming and expensive. With STAR-CCM+, the packing can be generated in a fast, physics-based and cost-efficient way using DEM. A completely automated workflow for packed bed simulation is possible, incorporating DEM, CFD and automation.

Figure 11 shows porosity values of non-spherical packings compared to experimental data for packings of spheres, cylinders, and cylinders with holes. DEM spheres are clustered to form a single element. Generation of the random packing is done with DEM and the positions of individual elements are fixed and exported. The geometry of the packed bed is created from the position and automated polyhedral meshing with proper contact resolution generates the computational mesh.

The simulation was run in STAR-CCM+ at a Reynolds Number of 6804 and wall temperature of 100°C. A sample comparison of heat transfer and porosity between STAR-CCM+ and experimental

Figure 9: STAR-CCM+ results from

Co-simulation in firebox side (left)

Figure 10: Process side results

showing temperature, heat flux

and species conversions (right)


80

data for the packed bed is shown in figure 12. STAR-CCM+ accurately predicts the packed bed physics (porosity, flow field and heat transfer) and enables reliable parameter design and optimization.

Reaction mechanism optimizationThe final part of the design puzzle is tuning the chemical reactions in the reactor. DARS (Digital Analysis of Reacting Systems) is a tool within STAR-CCM+ for 0D and 1D management and analyses of chemical reactions and modeling ideal reactors. Together with HEEDS, a global kinetic mechanism consisting of only a few reactions can be tuned to reproduce the results from a detailed (and accurate) reaction mechanism

before being used in STAR-CCM+ simulation. Since detailed reaction mechanisms in CFD are computationally expensive, this method can reduce the computational burden yet assure

the accuracy of a detailed reaction mechanism. With DARS and HEEDS, an optimized global mechanism can be developed in simple canonical systems that can be used in STAR-CCM+ at a computationally lower cost. A 12.5m packed bed reactor was analyzed with DARS at several H2O/CH4 ratios. Detailed and global chemistries are outlined in table 2. CO, H2, CH4 and temperature axial profiles are then transferred to HEEDS. The design objective in HEEDS is to minimize the root mean square (RMS) between the detailed chemistry reference curves and the global chemistry results. The chemistry parameters, pre-exponential factors, and activation energies were modified within the specified range by HEEDS and the DARS simulations

were run again. A total of 1,000 reaction designs were analyzed by HEEDS within two hours. A comparison of the baseline, detailed, and best designs are shown in figure 13. The

Figure 11: Individual components

of packed bed and randomly

generated bed from STAR-CCM+

Table 2: Detailed and global

chemistry outline for the profile

between STAR-CCM+ and

experimental data

reaction mechanism tuning

Detailed Chemistry [1] Global Chemistry [2]

Number of species 8 gaseous/13 Surface 5 species

Reactions 42 reactionsCH₄ +H₂O = CO = 3H₂CO + H₂O = CO₂ + H₂

CH₄ + 2H₂O = CO₂ = 4H₂


81

optimized reaction mechanisms were then used in the CFD simulations in place of the detailed chemistry, significantly reducing the computational cost with this automated workflow.

ConclusionWith CFD, a true data driven decision-making process while exploring design alternatives is possible for the design of reformers, heaters and furnaces. The combination of STAR-CCM+, DARS and HEEDS allows chemical engineers to simulate and optimize the performance and design of such equipment at the system and unit level. This multi-scale, multi-physics coupling approach can alleviate operational performance concerns while delivering improved reformer designs at reduced time and cost.

References1. Niveditha Krishnamoorthy, Nolan Halliday, Yuvraj Dewan and Ravindra Aglave: A CFD Model-Based Optimization of a Process Burner Geometry, AIChE Annual Meeting, Salt Lake City, Nov. 2015, Session 371a.2. Niveditha Krishnamoorthy, Thomas Eppinger and Ravindra Aglave: Multi-Scale Modeling of Fired Process Heaters in Chemical Process Industry, AIChE Spring Meeting, April 12-14, 2016, Session 109b.3. Thomas Eppinger, Nico Jurtz, Ravindra Aglave: Automate workflow for spatially resolved packed bed reactors with spherical and non-spherical particles, 10th International Conference on CFD in Oil & Gas, Metallurgical and Process Industries, SINTEF, Trondheim, Norway, June 17-19, 2014.

Figure 12: Comparison of heat

transfer and radial porosity (top)

Figure 13: Comparison of

baseline, detailed, and optimized

reaction chemistry (below)


82

83

Prime the pump | Life Sciences

Takehisa Mori - Terumo CorporationKuninori Masushige and Yuka Takahashi – Siemens PLM Software

The use of computer-aided engineering (CAE) significantly reduces workload, while ensuring reliability and improving medical device designs, an area where the development span is long and the authentication process is strict. Engineering simulation is now being widely recognized and is being adopted by regulatory authorities, manufacturers and suppliers around the world. One such example is the recent workshop on blood damage studies via modeling of a blood pump, organized by the US Food & Drug Administration (FDA), to set best practices and promote the use of simulation in device design. With increasing visibility and acceptance, CAE is now at the forefront of medical device design and is being actively deployed in bringing the latest products to market.

One such medical device is a Ventricular Assist Device (VAD), a mechanical pump used to increase blood flow to offset a malfunctioning ventricle in the heart and take over the function of the failing heart. While developing a VAD, computational modeling is an efficient tool in finding the best design of the blood pump, the key component, even before building a prototype. Blood pumps in VADs need to deliver the required blood flow while minimizing blood damage like hemolysis and thrombosis from the equipment. The

Prime the pumpIntroducing simulation-led design exploration to centrifugal blood pump development

use of CFD allows analysis and optimization of the blood flow and the pump, while offering quantitative prediction of hemolysis, recirculation and blood damage.

To talk about the recent developments in blood pump design using CFD and optimization techniques, we visited Terumo Corporation’s (hereinafter, “Terumo”) Shonan Center in Japan. The Shonan Center houses the R&D Center, where a variety of next-generation technology is developed, from the application of core technology to the development of new areas, such as diagnostic and therapeutic devices, along with myocardial regenerative therapy and device development for emerging markets.

Terumo - “Contributing to society through health care” Terumo was established in 1921 as a small-town factory near Tokyo to manufacture thermometers. The company was born from the desire of physicians to produce high-quality thermometers in Japan following a drastic reduction in the volume of thermometers imported from Germany due to World War I. One of the founders of the company was Dr. Shibasaburo Kitasato, who is known as the father of modern Japanese medicine and is famous for his discovery of the plague bacillus and his invention of

84

Life Sciences | Prime the pump

tetanus serum therapy. The company name originates from the German word for “thermometer.”

Based on the corporate mission of “contributing to society through healthcare,” Terumo is expanding its products and services to medical fields around the world through three business domains: cardiac and vascular, general hospitals, and blood management. Terumo is truly a global company with a high overseas sales ratio, with sales by region (as of March 31, 2015) at 63 percent abroad and 37 percent in Japan.

Numerical simulation at TerumoOn this occasion, we talked with Takehisa Mori of the R&D department’s exploration team. Mori is responsible for basic research on medical devices, along with investigation, design and development of next-generation products in the field of cardiovascular surgery. In particular, he evaluates the feasibility and challenges of new devices at the conception stage, utilizing computational fluid dynamics (CFD) analysis methods. The Exploration Team is not a specialized CAE department, but applies CFD and optimization techniques to design exploration. Within the field of cardiovascular surgical devices, CFD is used particularly in the development of blood pumps at Terumo. The purpose of introducing a CFD-based design exploration tool is to increase the efficiency of blood pump development and bring a better design to market, faster.

Terumo introduced CFD about ten years ago, and Mori was the first to apply it to

development. He assumed a leadership role in promoting CFD within the company and also served in a support role for other departments that were introducing it. In addition to blood pumps, there are still many areas at Terumo where CFD can be applied, such as artificial lungs, devices using drug solutions, pharmaceuticals, stents and various manufacturing processes.

“When I started with CFD, I realized how much I was able to understand designs better. I have been trying to share with researchers in other departments that the (information) range can be too narrow when using only experimentation, and that the range expands if you perform numerical analysis. One cautionary point is that some people think too lightly of simulation, assuming that anything can be simulated or imagined. In reality, it is important to consider what the problem means and what the physical implications are. If there are no such approaches or suppositions, the answers that come out of it are simply images that do not lead to the next step. It is possible that an incorrect solution will be used,” says Mori. “Especially in relation to medical devices, for which standards are strict, you can’t simply accept results - you must conduct substantial verification. Therefore, it can’t all be done with CFD. The prototype is also important. By utilizing CFD, we can build a base for prototype manufacturing to some extent.”

Application of CFD and optimization in blood pump developmentNext, we asked about design optimization using STAR-CCM+® software and Optimate+, the add-on for design exploration, in

Figure 1: Takehisa Mori, Principal

Research Manager, R&D Depart-

ment, Terumo Corporation

85


centrifugal blood pump design. Conventionally, the design method for blood pumps has involved using CFD to analyze dozens of initial-stage models in order to estimate the performance of each under the same conditions, and this method produced several candidate models. Then, pump performance evaluation and blood-related experiments were conducted, with problems in design being rectified. This process was repeated until a final prototype was determined. Consequently, the development period for one product took years.

With STAR-CCM+ and Optimate+, Terumo was able to significantly reduce the development time for blood pumps while achieving a better design, faster. Venturing into this optimization process required simulating a base design in STAR-CCM+ validated with experimental data, which could then be optimized using Optimate+. The simulations led to a prototype which underwent testing for validation of performance. The fluid used in the

experiments was blood, and because there are variations in the blood itself and variations in the experimental results, verification was difficult if a relatively long period of data was not available. Furthermore, the viscosity of blood varies from person to person. The blood used for the actual experiments came from animals such as cows, and viscosity differed depending on the type and age of the animal. Therefore, developers also experienced difficulties in accounting for the different viscosity parameters in the experiments.

Two examples of blood pumps used in medical settings are shown in figure 3. When designing a blood pump, it is necessary to meet the following requirements:> Low hemolysis: low blood trauma (hemolysis)> Antithrombogenicity: reduced risk of blood clots> Good pump efficiency: low power consumption, miniaturization of the drive system

Figure 2: Types of VAD: Extracor-

poreal-type CAPIOX® (left) and

Sarns® (right)

Figure 3: Blood circulation during

surgery (left); Implantable type:

left ventricular assist device

(right)

86


> Small and simple structure: portability, a size that is easy to implant, reduction in failure probability> Wide operating range: flow rate of 2-10 L/min; Lifting height of 50-800 mmHg

A streamlined optimization procedure is used because if a researcher is inexperienced, it is difficult to create an optimum form design without design optimization, and parameters that should be explored may be missed. Also, the goal is to see how efficient one can make the design process through optimization calculations.

Terumo researchers aimed at making an extracorporeal centrifugal pump with a general structure and improved pumping efficiency. The pump after optimization was as follows:> Number of impeller vanes: 12> Impeller diameter: 60 mm> Casing having a volute structure

The optimization settings used during the analysis are shown in table 1.

The pump efficiency (η) average value is characterized as: η ＝ (L × H) / (Rot × T)where L is the pump flow rate, Rot is the rotational speed (2,800 rpm), H is the

pump head (pressure loss) and T is the torque applied to the impeller.

As it was necessary to account for the respiratory rate range of an individual, the flow rate was changed in stages from 2 L/min to 8 L/min through a Java macro in STAR-CCM+. The pump efficiency and force applied to the impeller was output using the report function. The optimization calculation flow using Optimate+ is shown in figure 4.

The geometry of the pump was created using the 3D-CAD modeler in STAR-CCM+, as shown in figure 5. As 3D-CAD can parameterize the design variables and as additional licenses are unnecessary, optimization calculations can be implemented in a more seamless manner, saving time and additional license costs.

In addition, Optimate+ is a STAR-CCM+ add-on for the HEEDS™ multidisciplinary design exploration tool. Better designs can be found in a much smaller number of computations with SHERPA, the powerful exploration algorithm of HEEDS.

For the STAR-CCM+ analysis, blood was used as the working fluid, realizable k-ω model was used for the turbulence computations, and the rotational speed of the pump was set to 2,800 rpm. A polyhedral mesh was used, and a prism layer mesh was created for the wall, with approximately 3.4 million cells in total, as shown in figure 6.

The total number of analysis cases from the optimization process was 102, and the time required for the calculations was about six days. Based on the constraints and the base case, Optimate+ automatically refined the design variables (impeller diameter and casing) and iterated the simulations for all cases. Of these, 38 cases met the constraint conditions. Solutions that met all the optimization constraints are shown in figure 7, with the force on the impeller shown on the horizontal axis and the average pumping efficiency shown on the vertical axis.

The best design was defined, not as the one with the best global average pump efficiency, but as the one with the highest average efficiency and the least variations across the range of flow rates. Furthermore, the optimization results showed that a design with good pump efficiency that satisfies the constraint conditions tends to

Table 1: Summary of analysis

Figure 4: Workflow of the optimi-

zation calculations

87


be a design with small values for the height of the opening of the casing side and a small starting diameter. It was found that the shorter the interval between the impeller and casing is, the more efficient the pump becomes. The best design was identified within a week with the simulation-led optimization process.

Difficulties in pump designOne problem when designing a pump is that, although researchers want the pump to be efficient if efficiency takes too high a priority problems with hemolysis occur. Previously, the only way to assess the occurrence of thrombosis was through experimentation, but now a technique using CFD is being established at Terumo. The assessment method indicates that thrombi do not occur as easily if the shear rate in the retention area is lower than a certain threshold. Correlation with experiments is good. Therefore, it has gradually become possible to elucidate various phenomena using CFD instead of experiments. If a problem occurs with products designed before CFD application, it is possible to conduct a root cause analysis and change the design quickly by performing a re-analysis using CFD. Design and troubleshooting have become extremely efficient with the application of STAR-CCM+ to pump design. Previously, the structure of

medical devices was simple, thus it was possible to design such instruments without CFD, but due to the increase in design parameters associated with increased functionality, CFD and optimization calculations have become indispensable for design efficiency.

Lastly, Mori talked about important considerations in the implementation of optimization calculations: “A variety of solutions emerge when implementing optimization calculations. For pumps, values for efficiency etc., are output. However, at that time, if you use only these values, without confirming what is happening to the actual flow, it is possible that you will obtain a meaningless result a week later.

Figure 5: Geometry creation using 3D-CAD (left)

Figure 6: Illustration of mesh (below)

Figure 7: Analysis results show-

ing designs that satisfy the con-

straint conditions

88


Therefore, a few models should be made in advance, calculations performed, the flow checked properly, and then optimization calculations should be applied. This is more efficient. It is extremely important to check the flow field and understand it precisely.”

Reason for choosing STAR-CCM+STAR-CCM+ is equipped with 3D-CAD modeler, which enables the entire process, from geometry creation to postprocessing, to be implemented in a streamlined fashion from a single integrated user interface. It is especially attractive in that it does not require other software when implementing optimization calculations. In addition, meshing, including STAR-CCM+ surface wrapping for cleaning problematic geometry, is easy, says Mori. “These advantages are the reasons why the team chose STAR-CCM+. Previously, when we used different CFD software, we took the various components of the pump apart, meshing each part, and assembled it using the interface to create one analysis model. Then, we performed analysis, changed the model again, created the mesh, and repeated the process many times. Therefore, analysis was very time-consuming, and the meshing was extremely hard work. We also had to rely on another software to conduct the design optimization. With STAR-CCM+, the meshing is very simple, and the time required has been reduced significantly due to the in-built optimization capability. We basically wanted our workforce to focus more on product design, thus spending a lot of workforce resources

for computational modeling is like putting the cart before the horse. This has been improved tremendously.”

The next challenge in further fine-tuning this process is to create guidelines for best practices for meshing and analysis. The product groups developed at Terumo are nearly one-of-a-kind, such as their blood pumps and infusion devices, and there is a real need to create standardized best practices for simulation.

ConclusionMori is very satisfied with the technical support from Siemens PLM Software, which has helped the group achieve success in deploying simulation-led design. The “My Case” feature and knowledge base articles (FAQ) of the Steve Portal, and Support News, an e-mail magazine, are helpful in directly connecting with the support team at Siemens PLM Software to resolve issues. Technical workshops are also utilized for CFD expansion at Terumo and are contributing to the increase in actual users within the company.

Furthermore, because the overseas ratio of Terumo sales is very high, medical devices are not only being established for use in the Japanese market exclusively, but products are also designed with the global market in mind, from the start of planning and design. Through their use of STAR-CCM+ and Optimate+, Terumo can now discover better designs, faster and play a part in “contributing to society through health care.”

Figure 8: Analysis results show-

ing the pressure profile of the

best design

89

90

Figure 1: Constriction in the airways of an elderly patient with goiter

91

Case studyAn elderly lady comes into a west London hospital complaining of breathing difficulties. She has a large growth on her neck (a goiter, an enlargement of the thyroid gland) and this is compressing her trachea (windpipe). However, she also has an underlying lung problem. The important questions her doctors need to answer are: First, what is causing her breathing issues, her lungs or the growth pressing on her windpipe? Then, is it worth performing a potentially risky operation on an elderly lady to remove the goiter if the problem is due to her lungs?

Currently, surgeons do not possess the tools to assess the relative contributions of the large airways and lungs to the problems of breathing. Is this a role in which computational fluid dynamics (CFD) could provide key diagnostic help in hospitals? Can CFD provide a tool that allows the airways to be considered separately from the lungs?

Ten months pass and our patient has deteriorated. She is now wheelchair-bound, purely due to her breathing. The “work of breathing” is the amount of effort it takes a person to draw their next breath – to expand their lungs and overcome the resistance to air flowing from the outside down to the alveoli. In our patient, this “work of breathing” has increased so much that all the energy she gets from the oxygen in one breath is used up drawing the next; she has no surplus energy with which to walk around or be active in any way. It represents a huge deterioration in her quality of life.

The value of CFD in respiratory medicine

Doctors can use computed tomography (CT) scans to assess the impact the goiter has on the trachea shape. Our patient had been scanned twice, first when she initially came to the hospital and then when she returned 10 months later. Whilst there was evidence of some narrowing found over the 10-month period, it did not seem severe enough to cause the dramatic deterioration in the patient’s breathing.

In order to gain some insight into what was happening with the patient’s airflow, the surgeons turned to their collaborators at the Department of Aeronautics at Imperial College London. CFD simulations using STAR-CCM+® software were performed based on the anatomy seen in the two CT scans. The increase in airway resistance was seen to rise threefold during the period over which the patient had deteriorated and was 15 times more than what might be expected in a healthy subject breathing quietly. The upper graph in figure 1 shows the cross sectional area variation along the trachea in May 2012, April 2013 and post surgery. A small difference in the narrowest point of the trachea can be seen between May 2012 and April 2013. This is the cause of significant pressure losses, as illustrated on the lower graph. Figure 2 shows the total pressure on the airway surface, facing forward and sideways. Given the evidence that the changing anatomy was causing such an increase in the patient’s work of breathing, the surgeons made the decision to operate to remove the goiter and allow the trachea to return to its normal shape.

Alister Bates - Imperial College London

The value of CFD in respiratory medicine | Life Sciences

92

The operation was performed successfully, but would the outcome follow the CFD prediction or was the problem in the lungs all along?

The answer could not have been more emphatic; our patient quickly recovered and was restored to a greater quality of life than she had enjoyed even before her initial hospital visit. The insight provided by the CFD allowed her condition to be accurately assessed and the skill of the surgeon restored her to health.

CFD simulations were performed based on her postoperative imaging and confirmed that the resistance to airflow had reduced from the pre-op case to normal levels. The change in airflow patterns between the pre- and post-operative cases is shown in figure 3. Before the operation, the constriction caused very fast moving flow, which

separated downstream of the constriction, causing a large recirculation region. Post-op, the flow is much slower through the trachea and far less unsteady.

So why had the resistance to airflow increased so much over the 10-month period? Current assessment of goiters is based on the reduction in airway diameter between a healthy segment of trachea and the narrowest point. In our patient, this had not changed significantly, but the goiter had caused the trachea to narrow along a far greater length. The shape of the constriction had also changed; consequently, as the flow passed through the constriction, it now separated away from the airway wall. A significant portion of the downstream trachea was taken up with recirculation of the air, forcing the useful flow of air into an even narrower jet than would be expected simply from the geometry alone.

Figure 2:

Side view of the airway during

the second pre-op scan (April

2013), with the subject in supine

position. A volume rendering of

the airflow velocity is shown to

allow comparison between the

flow patterns and measurements

shown in the two graphs under-

neath. (top)

Cross-sectional area of the tra-

chea at three instances in time

plotted from the first tracheal

ring to where the trachea bifur-

cates into the bronchi. In each

case the cross-sectional area is

normalized by the area at the

first tracheal ring. The constric-

tion affecting the May 2012 and

April 2013 cases occurs 20-30 mm

below the first tracheal ring. The

large change in area between the

pre- and post-op cases shows the

effect the surgery had on the

airway anatomy. (middle)

Mean total pressure loss in the

descending airway over the same

region of anatomy as above. The

curves values at 0 mm (the first

tracheal ring) show pressure

losses as flow passes through the

glottis into the trachea, with 0 Pa

relative pressure equaling atmo-

spheric pressure. In the post-op-

erative case, no subsequent loss

is observable on this scale, whilst

the pre-op cases both exhibit

large losses through the con-

stricted region. The May 2012

and April 2013 cases show mark-

edly different pressure loss

curves despite similar area distri-

butions in the graph above.

(below)

Life Sciences | The value of CFD in respiratory medicine

93

This case shows the potential for CFD to provide important answers to difficult clinical questions. Whilst further work is needed to capture the full physics involved in breathing, it is already clear that useful information can be provided to guide surgical decision-making and improve the quality of life in those suffering from airway disease.

Performing respiratory CFDThe airways carry out many roles besides just acting as the conduit for air to reach the depths of the lungs. The nose filters out potentially dangerous particles and vapors while warming and heating the air while also providing our sense of smell. CFD is providing us with insights into how our airways control the flow of air within them, as well as how this can be exploited to target drug delivery.

Performing CFD in the body follows the same steps as any other simulation: geometry creation, meshing, solving, postprocessing and validation. Using medical images, such as either CT or magnetic resonance (MRI) as a starting point, the first step is segmentation – the process of delineating which voxels (3D pixels) in the medical image represent the airways. The edges of these selected voxels therefore represent the surface of the

airway walls, which need smoothing to more faithfully represent the real anatomy.

The choice of boundary conditions and solver parameters is critical. For instance, internal pressures are difficult to measure in people, as are other quantities such as air temperature and humidity and airway wall stiffness. The easiest variable to drive the flow with is the rate of flow outside the face, as it can be directly measured in a subject. Applying this flow rate far from the face allows the flow to develop naturally at the entrance to the mouth or nose. The pressure loss along the airways can then be calculated in the simulation and converted to a clinically relevant measure, in this case the large airway’s contribution to the “work of breathing.”

Furthermore, breathing is a reciprocating process with unsteady flows both in and out, making simple steady flow simulations unrepresentative of the real life situation. Fortunately, previous studies have shown that mean flow patterns are constant for 90 percent of an inhalation, even though the flow rate changes¹, meaning that quasi-steady simulations (unsteady simulations with constant boundary conditions) are a good compromise between the expense of simulating a full breath and obtaining sufficiently accurate and informative results.

Figure 3: Two views of the airway

surface shown from the pharynx

to the bottom of the trachea dur-

ing the second pre-op scan (April

2013). The total pressure on the

surface is displayed. The airway

is shown within a volume render-

ing of the subject’s skeleton to

aid in locating the anatomy. An

abrupt drop in total pressure is

apparent through the constric-

tion at the level of the collar-

bone. (top)

Figure 4: The airflow velocity

magnitudes in April 2013 and

post-op are shown. The pre-op

case shows a high velocity region

through the constriction that is

exacerbated as this region pro-

duces a jet that impinges on the

tracheal wall, forcing high veloci-

ty flow to one side of the trachea

and preventing redistribution of

the airflow. The effect of remov-

ing the goiter can be seen in the

lack of high velocity region in the

post-op case. (below)

The value of CFD in respiratory medicine | Life Sciences

94

The airway geometry is extremely complex, starting with the nasal and oral entrances and descending through various constrictions in the pharynx and larynx before entering the trachea through the vocal folds. This complex anatomy introduces turbulence at Reynolds numbers far below what may be expected in simple pipes. In order to capture this turbulence, LES simulations were employed for the case study described above, utilizing around 2.2 million elements for each airway. Each mesh comprised of five prismatic layers, with polyhedral cells making up the bulk of the mesh. The temporal discretization is also critical in capturing the flow dynamics within an inhalation and was set to 0.1 ms in these simulations. These simulations were validated by comparison to experiments performed on 3D printed tracheas. Further validation was performed by comparison to very high resolution (near DNS) simulations with length and time scales approaching that of the Kolmogorov scales.

About usOur group at Imperial College London has been involved in many studies involving CFD within the airways. As well as more in-depth investigations into the effect of goiters2, 3, as in the case study above, we have analyzed the world’s first pediatric transplanted trachea and shown how the patient’s breathing effort changed as he grew with the transplanted segment⁴. We have used particle models in STAR-CCM+ to investigate deposition during sniffing, which can help to improve nasal drug delivery techniques¹. We are also currently moving towards studying increasingly complex physiology through modeling

BiographyAlister is currently a Research Fellow at Cincinnati Children’s Hospital where he is using CFD to enhance the treat-ment of children with sleep apnea. The research presented here was per-formed during his PhD and fellowship at the Department of Aeronautics, Im-perial College London with Professor Doorly. He has also worked as a CFD aerodynamicist for the Williams For-mula 1 team and consults for the bio-engineering firm Anatasys. He will re-turn to Imperial College in 2017.

References

1. Bates, A. J. et al: Dynamics of

airflow in a short inhalation. J. R.

Soc. Interface 12, 20140880–

20140880 (2014).

2. Bates, A. J. et al. Power loss

mechanisms in pathological

tracheas. J. Biomech. (2015).

doi:10.1016/j.

jbiomech.2015.11.033.

3. Bates, A.J. et al. The Effects of

Curvature and Constriction on

Airflow and Energy Loss in

Pathological Tracheas. Respir.

Physiol. Neurobiol.

4. Hamilton, N. J. et al. Tissue-

Engineered Tracheal

Replacement in a Child: A 4-Year

Follow-Up Study. Am J Transpl.

(2015). doi:10.1111/ajt.13318.

dynamic airway wall motion and heat and water transfer in the large airways.

The teamDr. Alister Bates - Upper Airway Center, Cincinnati Children’s Hospital, USA; Department of Bioengineering, Imperial College London, UKDr. Andrew Comerford - Imaging Sciences & Biomedical Engineering, King’s College London, UKProf. Denis Doorly - Department of Aeronautics, Imperial College London, UKDr. Raul Cetto - Department of Aeronautics, Imperial College London, UK; Department of Otolaryngology, St. Mary’s Hospital, Imperial College Healthcare Trust, UKDr. Neil Tolley - Department of Otolaryngology, St. Mary’s Hospital, Imperial College Healthcare Trust, UKProf. Bob Schroter - Department of Bioengineering, Imperial College London, UK

“The insight provided by the CFD allowed the patient’s condition to be accurately

assessed and the skill of the surgeon restored her to health.”

Life Sciences | The value of CFD in respiratory medicine

95



siemens.com/mdx

96

97

James Brett - Synergetics

Figure 1: Schematic of the

bottom surface of the stepped

water channel. The section

highlighted in blue was used in

the STAR-CCM+ model. A periodic

interface was used to map the

outlet flow parameters back to

the inlet.

IntroductionIn the water treatment industry, managing of metrics such as pH, additive concentration and micro-scale organisms is essential for assuring quality control over treatment processes. For facilities that process contaminated waste water, proper treatment is essential for achieving hygiene, safety and environmental goals.Computational fluid dynamics (CFD) is a valuable tool for understanding water treatment processes and is especially valuable for modelling scenarios that cannot readily be tested in situ. In this article, we analyze mixing effects along a stepped channel. This analysis uses several STAR-CCM+® Software features, including: unsteady flow solver, multiphase modelling, dynamic table based mesh refinement and passive scalar transport.

The problemWater treatment engineers were planning to modify an existing treatment process by introducing an additive into water flowing along a stepped channel. The channel consists of five stepped weirs with each weir separated by a distance of 40 m, as illustrated in figure 1. Each weir is 2 m higher than the channel bottom and each

Modeling additive mixing in weir flows

step drops the channel depth by 2 m. Due to the layout of the treatment facility, it was most practical to insert the additive at the end of the channel. However, it was also a requirement that the additive be well mixed into the flow before exiting the channel. The aim of this study was to answer the question: how many weirs must the additive/water mixture pass over for the additive to become well mixed into the flow? Answering this question would enable water treatment engineers to optimally locate the additive insertion point.

Modeling methodWe modeled this additive mixing process using STAR-CCM+® Software. To reduce computational requirements, only one weir was modeled (the section highlighted in blue in figure 1). A periodic internal interface was used to map the outlet flow back to the inlet. In effect, this interfaced approach models an infinitely long channel with an infinite number of stepped weirs. This approach reduced the total domain size and mesh size by a factor of five, significantly reducing computation time.We used an Eulerian multiphase volume of fluid approach to model the water-air phase interactions. The flow field was initialised

40 m

40 m

Modeling additive mixing in weir flows | Environment

98

with zero velocity and 100 m3 of water within the modelled section of the channel, as shown in figure 2. Turbulence was modelled using a standard k-ε model and two-layer all y+ wall treatment was used to resolve the boundary layer at the channel walls. We employed a second order unsteady model with a time step size of 0.01 s, to adequately resolve the dynamic flow structures.

An example volume mesh along a cut plane through the channel is shown in figure 3. We used the trimmed mesh, in conjunction with dynamic mesh refinement. This automatically refined the mesh at the water surface and in water regions with high air entrainment, using a table refinement. A Java macro was used to extract the table and perform a re-mesh every 20 time steps.

To model the additive dispersion and mixing along the channel we added a passive scalar immediately above the weir. This was initialised within a cubic volume with a width of 0.25 m. The flow was allowed time to develop before the additive was introduced into the water phase. The mixing quality, MQ, was assessed using a non-dimensionalised parameter,

where Ci is the concentration in a cell, C is the average concentration in the water phase, Vi is the volume of water in a cell, and V is the total volume of water. This formulation tends towards a mixing quality of zero when additive is poorly mixed, and is one for a perfectly mixed solution. For this analysis, a mixing quality of 0.5 was sufficient for the additive to be considered well mixed into the flow.

ResultsFlow over the weir was modelled for 32.5 s after the additive was injected into the flow. Velocity at the water-air free surface is shown in figure 4. Velocity at the surface is highest as water flows over the weir. Immediately after the weir, a large recirculation region forms with high levels of air entrainment.The change in additive concentration with time is easily visualised in STAR-CCM+: figure 5 shows how the additive passes over the weir and how the periodic interface maps the outlet concentration back to the inlet. Figure 6 shows how the mixing

Periodic interface

Figure 2: Section plane through

the water channel showing initial

volume fraction of water phase

and the location of the periodic

interfaces.

(top)

Figure 3: Section plane along the

channel showing mesh

refinement at the water-air

interface and in areas of mixing

between the phases.

(below)

Environment | Modeling additive mixing in weir flows

99

0s

1s

2s

5s

10s

30s

quality, MQ, develops with time. It takes 32.5 s for MQ to reach 0.5, meaning the additive is well mixed into the flow. During this time, the additive front passes over the weir three times. This result demonstrates that the additive must be inserted into the channel at least three weirs upstream of the exit in order to become well mixed.

ConclusionCFD is a valuable tool for solving water treatment flow problems, particularly at treatment facilities where in situ testing is not feasible. In this study, STAR-CCM+ enabled us to model additive mixing along

Figure 4: Flow velocity on the

water-air interface. Highest flow

velocity occurs as water flows

over the wei (top)

Figure 5: Passive scalar

concentration at various time

steps. The periodic interface

maps the downstream

concentration back to the

upstream boundary (below left)

Figure 6: Mixing quality versus

time (below right)

Modeling additive mixing in weir flows | Environment

a narrow channel containing five stepped weirs, and determine the optimal additive location to ensure well mixed flow at the end of the channel. This model used a simplified domain with a single weir and periodic interfaces. The unsteady multiphase model and table based mesh refinement enabled us to resolve the flow field at the water surface and air entrainment and the additive concentration has been modelled using passive scalar transport. From the modelling we have determined that the additive should pass over at least three weirs in order to become well mixed into the flow.

100

101

Christopher Beves - Siemens PLM Software

The “Special Relationship” was a phrase coined by Sir Winston Churchill to describe the close ties between the U.S. and the U.K. throughout the events of the 20th century. This same unique “Special Relationship” arguably exists between STAR-CCM+® Software users and their Dedicated Support Engineers (DSEs), particularly come 4 p.m. on a Friday afternoon when a client needs results for first thing Monday morning and help is needed! On a recent project, the CFD modeling team at WSP Parsons Brinckerhoff wanted to exploit the Solid Stress capability within STAR-CCM+ in order to assess a potential Fluid-Structure Interaction problem. Without having the desired development time required to test and implement this, they utilized the “Special Relationship” with their Designated Support Engineer to assist them. I sat down with Associate Director James Bertwistle and Associate Andrew Basford during one of the DSE support visits to WSP to hear about their work, the project in question and how they overcame the challenge as it arose.

Bertwistle, having single-handedly built the team from the ground up to a team of

When the wind blows…

five, introduces us to who they are: “The CFD team exists as a specialist discipline within WSP Parsons Brinckerhoff, which is an international multidisciplinary engineering consultancy and has offices spanning five continents with about 35,000 staff. The core services of the company are mechanical engineering, structural engineering and environmental planning. The CFD team members work as internal consultants to support all of these disciplines with whatever simulation requirements they might have, and to assist external clients as well.”

Being part of such a large engineering organization working on varied projects may lead to being pulled in many different directions at once, such as fire and smoke modeling, pollutant dispersion and thermal comfort studies. However, as James explains “We end up providing a diverse range of support, but one of our main CFD services is pedestrian wind simulation, which falls within the broader scope of Computational Wind Engineering. For our pedestrian wind cases this would involve helping architects as they go through the planning process for new buildings and to assess pedestrian safety

Figure 1: CFD modeling team

from left to right: Rama

Pathakota, Andrew Basford,

Joshua Millar, James Bertwistle

When the wind blows... | Support

102

when a new development pops up in a large urban environment. The focus on safety inherently raises the level of risk on our projects to the client.

“We use CFD where the wind tunnel might not be the quickest method to assess multiple prototype wind mitigation solutions, and to give us greater understanding from the whole field data as to what wind directions or local phenomena are causing any issues. Atmospheric Boundary Layer modeling is complex and our work is consistent to the European COST Action C14 framework for Computational Wind Engineering, and we have been fortunate lately to devote some research and development time into validation of our CFD methodology with wind tunnel data at Cranfield University. Looking to the future, we are investigating services we can develop using an unsteady approach to let us capture transient gusts and open up additional capabilities including Fluid-Structure Interaction and dynamic building response for our structural engineering team.”

As Fluid-structure interaction is something they have been interested in developing for a while, it was only a matter of time until a project came up that allowed them to look into it as fellow team member, Associate Andrew Basford touches on: “On a recent job, we had an issue when we were looking at a taller building compared to its mostly lower level surroundings with high velocities at ground level. The solution to this is traditionally a glass/steel canopy over the entrance to shield the pedestrians from the high velocity downwash. However, the structural team (was) concerned with the potential pressures this canopy would be subjected to in a storm scenario - Would it break? Would it flutter and tear itself apart?

Traditionally, a lot of the work in terms of wind pressure loading was done using building codes, but by using CFD and in particular FSI we can use more performance-based methods to analyze these problems.”

The traditional wind tunnel approach, while an industry stalwart, is also not without its limitations, as James explains: “These thin, composite structures are difficult to analyze in a wind tunnel because of scaling and instrumentation issues, expensive physical models and notoriously long lead times required to get into a wind tunnel. But by offering a CFD-based approach this would provide us with more rapid insight and detail of the building façade pressure loadings.”

As is the usual response by engineers needing to find a solution in a pinch, the simplest solution was tested first. Andrew Basford: “Being so early on in the design phase it was clear that wind tunnel studies were out of the question. It was easy enough for us to model the canopy as a static part of the building in CAD and get the pressure from that. But that wasn’t going to answer the question, ‘Will it flutter? ’. By conducting an FSI simulation in STAR-CCM+, which could be done in a few hours, that would enable us to get a much greater level of understanding early in the design process when ideas are being put on the table by being able to account for the glass and steel structure together, and reduce risk moving forward. Being so tight on time, as consultancy generally is, we contacted our local DSE in the London office to give us some best practice guidance on the set up so that we could model it with confidence the first time before going into a meeting with the results.”By logging a support request to their local

Figure 2: WSP Parsons

Brinckerhoff projects around the

world: a) The Shard (London);

b) World Trade Center (New York);

c) Zayed National Museum (Abu

Dhabi: image courtesy of WSP

Parsons Brinckerhoff)

Support | When the wind blows...

103

DSE in London, Annabella Grozescu, the “Special Relationship” between STAR-CCM+ user and their DSE was called into play. They were able to utilize the unique one on one support system to help them tackle this problem. Annabella explains: “I support about 10 companies. Whenever a customer has a problem, they can contact me directly. It ’s easier for me to understand their needs because I already know who they are, their applications, and what types of problems they are working on. I am in regular contact with my customers. If I don’t hear anything from them for a while, I will contact them to see if they’re successfully applying new features or methods, if they are having any problems, or just to see how they’re doing.”

After getting the required information from James and Andrew, Annabella was able to set up an FSI simulation of the glass and steel canopy on the building façade and account for the expected storm force speeds. Annabella explains, “The fully coupled FSI simulation, with the steel frame fixed to the ground, showed some static deflections of the glass but there was no vibration problem. I was able to feed that simulation setup back to James and Andrew. On one of our scheduled support visits, I was able to follow it up with them in person to see how they got on with it.”

James reflects on the effectiveness of going through the DSE on this issue: “Getting the set-up information helped us a lot. Buildings take a long time to be built, so in the early concept phases of a design we can give architects and our specialist design teams more information about how a building is going to perform. We find working in this collaborative way reduces our development risk. It gives us access to

the brains behind the coding that we are using so we can really understand that we are using it correctly and we are not misapplying it, which could potentially have disastrous effects for our clients, our client projects, our client safety, and our own indemnity insurance.”

In summing up the collaborative way that WSP Parsons Brinckerhoff uses their DSE, Andrew adds, “We have been using STAR-CCM+ for over four years now. Initially our questions were simple things about get-ting to know the software. But as we have become more experienced now it ’s looking at trying to ensure that when new features come out we can implement them in the most accurate and efficient way. Having the same point of contact as a DSE makes it easier for us to develop new methods and have an extra opinion on a set-up is-sue if needed.”

Given that they now have access to additional modeling capabilities, it would be interesting to get their thoughts on what they think this means for their team. Andrew Basford makes the point: “With the complexity of buildings increasing, it ’s becoming more important to ensure we can carry out the most accurate analyses possible. The way analyses in CFD are going, it is much more about how a whole system works and not just a single abstract or problem.” James expands further on the growth of their increased modeling options: “Now that we understand how to set up a multi-component structure in a fully coupled FSI simulation, it will allow us to understand what design freedoms this will unlock for our clients, what limitations they might have and how we can best apply them to meet the needs of our clients and add value to their projects. Moving forward it will allow us to bid for and win more work which is always exciting.”

Figure 3: Andrew Basford and

Annabella Grozescu on the

support visit (left)

Figure 4: Glass canopy deflection

(right)

When the wind blows... | Support

104

STAR-CCM+ global academic program.Educating tomorrow's engineers today.

siemens.com/mdx

About Siemens PLM Software Siemens PLM Software, a business unit of the Siemens Digital Factory Division, is a leading global provider of product lifecycle management (PLM) and manufacturing operations management (MOM) software, systems and services with over 15 million licensed seats and more than 140,000 customers worldwide. Headquartered in Plano, Texas, Siemens PLM Software works collaboratively with its customers to provide industry software solutions that help companies everywhere achieve a sustainable competitive advantage by making real the innovations that matter. For more information on Siemens PLM Software products and services, visit www.siemens.com/plm.

Headquarters Granite Park One 5800 Granite Parkway Suite 600 Plano, TX 75024 USA +1 972 987 3000

Americas Granite Park One 5800 Granite Parkway Suite 600 Plano, TX 75024 USA +1 314 264 8499

Europe Stephenson House Sir William Siemens Square Frimley, Camberley Surrey, GU16 8QD +44 (0) 1276 413200

Asia-PacificSuites 4301-4302, 43/F AIA Kowloon Tower, Landmark East 100 How Ming Street Kwun Tong, Kowloon Hong Kong +852 2230 3308

© 2017 Siemens Product Lifecycle Management Software Inc. Siemens and the Siemens logo are registered trademarks of Siemens AG. STAR-CCM+ is a trademark or registered trademark of Computational Dynamics Limited. HEEDS is a trademark or registered trademark of Red Cedar Technology, Inc. All other trademarks, registered trade-marks or service marks belong to their respective holders.

Documents

Designed with STAR-CCM+: The world’s best bottles. · From the blog: Flo Cycling wheels 4-5 From the blog: Question and learn with data focus 6-7 Bottero: Stronger, lighter glass