Team G13

Generators: Aditya Prasad, Shivaji Yadav

Ideators: Piyush Anand, Darshan Bitla

Planners: Naufran Neyas, Ramresh Meena

Solvers: Krishna gaggar, Pranjal Sankhwar

 Table of contents
1. User Story

2. Technical Problem Catered

3. Introduction
1. Data center physical organization and overview
2. Data center thermal loads and temperature limits

4. Simplified Geometry

5. Control Volume

6. 1-D Temperature Profile

7. Known and unknown factors

8. Three Modes of Estimation

9. General Assumption

10. Solving Methodology

11. Waste heat recovery

1. Hot water production
2. District heating
3. Power Plant
4. Organic Rankine Cycle
5. Absorption cooling

12. Conclusion

13. References
 User Story

A revived interest in the collection and reuse of waste energy is being driven in part by the
depletion of the planet's finite reserves of fossil fuels, the global warming crisis, and the high
cost of energy. As the need for cloud-based connection and performance rises, data centers are
becoming a significant source of energy waste. In fact, according to current data, data centers
account for more than 2% of all power used in the US, and this is predicted to reach 15-20 % by
2030. This electricity is almost entirely used to cool the electronics, which generates a sizable
amount of waste heat. Due to the low quality of the heat, it is challenging to recover and reuse
this stream of waste heat. In this report, the most promising approaches and technologies for
efficiently and economically recovering low-grade waste heat from data centers are identified
and discussed.

Technical Problem Catered

Part 1 -

Consider a rack with dimensions 78 in x 28 in x 24 in, containing 2184 pipes. Each pipe has a
dimension of 1.41 mm x 11.6 mm x 28 in and 52 of them correspond to a server or 1U. The rack
is simplified into a uniform box with a temperature of 25 C. Water-glycol mixture is flowing
through the 42 pipes acting as heat sink for cooling the rack. The Objective is to find temperature
difference between inlet and outlet so as to omptimize waste heat production and utilization.

Part 2 -

The suitability of popular waste heat recovery strategies for low temperature, high volume waste
heat generation is examined using the waste heat generated. District/plant/water heating, direct
power generation (thermoelectri), indirect power production (organic Rankine cycle and steam),
are a few of the recovery strategies. In order to evaluate each technology's applicability and
efficacy for data centre applications, these methods are assessed together with their operational
needs.
 Introduction

1.Data center physical organization and overview

The majority of ICT (Information and Communication Technology) modular assets, including
servers, switches, and storage facilities, are housed in a data center, which also controls the
environmental factors (temperature, humidity, and dust) to ensure that the ICT systems function
dependably, safely, and efficiently. A data center may contain a single rack of equipment, a
couple, or even many racks and cabinets, depending on the size of the company. A rack is a
metal frame or container that is standardized and into which ICT modular components are fitted
horizontally. A typical rack measures 78 inches in height, 23 to 25 inches in breadth, and 26 to
30 inches in depth. The unit of measurement "U" is used to define the height or thickness of a
modular asset installed in a rack. U about translates to 1.8 inches. The majority of servers (often
those with one or two socket processors) are one unit thick (U), although some bigger servers
(such those with four socket multiprocessors) may be two U or more thick. A typical full-size
rack may hold 42 modular 1 U assets in total.
2.Data center thermal loads and temperature limits

Manufacturers are designing and producing more compact and higher power modules as a result
of the rising demand for ICT services on the one hand and the direct correlation between data
center expenses and floor area on the other. The energy flux dissipated by older generations of
data centers has grown at least ten times (from 6458 to 10,764 W/m2), whereas the energy flux
dissipated by traditional data centers is in the range of 430 to 861 W/m2. The design and
production of thermal management systems is one of the most difficult parts of data center
architecture when compared to the capability of traditional HVAC systems for rooms of a
comparable size (40-86 W/m2).
While in traditional data centers, per-rack power consumption is around 7 kW, a high
performance fully used rack's power consumption is in the range of 10-15 kW, and racks stacked
with blade servers may dissipate up to 21 kW of electricity.

For the safe and efficient operation of microprocessors, 85 °C is generally accepted as the
highest permissible junction temperature in electronics thermal management research. The
literature does contain a few more references, though, that suggest somewhat higher or lower
ranges as the temperature limit. For instance, it is thought that 78 to 100 °C is the maximum
working temperature for microprocessors. The same 85 °C maximum temperature is often
applied to DIMMs as well. Hard disc drive temperature limits are substantially lower than those
for microprocessors and DIMMs, though. In reality, operating at temperatures exceeding 40–45
°C for extended periods of time raises the risk of HDD failure. For their product disc drives,
several manufacturers have still established greater temperature restrictions of up to 60 °C.

Simplified Geometry
Control Volume

1-D Temperature Profile

Known Factors:
We know average power/heat generated by a rack, demand Temperatures by different waste heat
utilization technologies, Geometry of control volume.

Unknown factors:
Water mass flow rate, Inlet temperature, Surface temperature of rack,

Three modes of estimation:

1. Conduction
We have water flowing in pipes acting as a heat sink for rack, the pipe walls will conduct
heat from processors to water flowing inside. This will have a heat transfer coefficient
associated with it which we can find using resistance theory.

2. Radiation
The temperatures are low so radiation won’t have an appreciable effect. (In our midterm
report we got radiation of order 100 W which is negligible compared to convectional and
conduction components.)

3. Convection
Convection is the main form of heat transfer through which we are extracting waste heat.
The associated heat transfer coefficient can be calculated using Nusselt number.

General assumptions:

1. Radiation is neglected in analysis.

2. The rack is modelled as a constant temperature surface of 75 °C.

3. The flow is assumed to be fully developed.

4. Flow of water is in steady state, no temporal variation.

5. Flow resistance in pipe is neglected in analysis.

6. Pipe is insulated on its vertical walls, negating any air convection effects.

7. All fluid properties are taken at median temperature of operation

Approximate estimates:
Solution Methodology:
 Waste heat recovery
In this section we review eight potential waste heat recovery technologies and their temperature
demands. We find out for each method what is the operating temperature conditions (inlet
temperature, outlet temperature and surface temperature for data center) and efficiency achieved.

1.Hot water production

HVAC or hot water production systems are two frequent and reasonably easy
applications for the reuse of low-quality energy. The server waste heat may be recovered
at temperatures between 35 and 45 °C, which is more than enough to be used for heating
purposes. Depending on the size and operating conditions of the data center, the space
heating provided by data centers can range from the HVAC needs of the data center itself
to that of a single-family home, an apartment complex, a neighborhood or even several
neighborhoods (district heating). The use of remotely siting servers in individual homes
to provide domestic heating is referred as a “Data Furnace”.

Based on this temperature demand (35 - 45 °C), applying our solution methodology we
get,
Inlet temperature = 25 °C
Outlet temperature = 35 - 45°C
Surface temperature = 36 - 47 °C
Efficiency = 6.04 – 12.08 %

2.District heating
Another popular low-quality waste heat recovery technique that is both economically and
environmentally advantageous is district heating. By using liquid cooling, it is possible to
capture waste heat with a somewhat higher quality (up to 50–60 °C as opposed to 35–45
°C), which may then be utilized to heat a larger area. Thus, the data center operator may
benefit financially from this heat.

Based on this temperature demand (50 - 60 °C), applying our solution methodology we
get,
Inlet temperature = 25 °C
Outlet temperature = 50 - 60 °C
Surface temperature = 52.6 – 63.66 °C
Efficiency = 15.1 – 21.1 %
3.Power Plant

The use of waste heat to provide heating of water in the thermal Rankine cycle of a major
power plant is a common waste heat recovery technique. The most effective waste heat
source for power plant boiler feed-water preheating is better quality waste heat than is
typically provided by data centers. The normal minimum requirement is between 60 and
100 °C, which is the maximum amount of waste heat from water-cooled data centers that
is accessible. The advantages of employing data center waste heat for boiler feed-water
preheating in power plants include potential revenue from selling heat to the power plant
as well as potential carbon offsets for the data center operator.

Based on this temperature demand (60 - 100 °C), applying our solution methodology we
get,
Inlet temperature = 45 °C
Outlet temperature = 60 - 80 °C
Surface temperature = 61.57 – 83.67 °C
Efficiency = 9.06 – 21.14 %
4.Organic Rankine Cycle

An organic Rankine cycle (ORC) may be utilized to directly generate power from data
center waste heat. Similar to the steam Rankine cycle, organic fluids with much lower
boiling temperatures are used as the working fluid in ORCs. The thermodynamic
properties of the working fluid strongly influence cycle efficiency and ORCs can operate
successfully with waste heat streams of 65 °C and higher and can even run as low as 32
°C with reduced efficiency. The advantages of employing data center waste heat in
organic Rankine cycle power generation include the on-site energy production from
waste heat and the lack of any particular siting requirements, making it suitable for a
variety of data center designs.

Based on this temperature demand (32 - 65°C), applying our solution methodology we
get,
Inlet temperature = 25 °C
Outlet temperature = 32 - 65 °C
Surface temperature = 32.73 – 69.11 °C
Efficiency = 4.3 - 24.16 %
5.Absorption cooling

Generator temperatures of 70 to 90 °C, which are commensurate with the available waste
heat from water-cooled, are sufficient for absorption refrigeration systems to function. By
producing chilled water for chilling, the use of data center waste heat in absorption
refrigeration systems directly reduces the load on data center CRAC systems, which is
advantageous economically for the majority of data center owners. Furthermore, there are
no concerns with site co-location, and if there is enough room, the technology may be
retrofitted into existing data centers.

Based on this temperature demand (70 - 90°C), applying our solution methodology we
get,
Inlet temperature = 50 °C
Outlet temperature = 70 - 90 °C
Surface temperature = 72.1– 94.19 °C
Efficiency = 12.1 - 24.16 %
 Waste recovery methods Summary:

Waste Recovery Inlet Temperature Outlet Temperature Surface Efficiency

Technique Temperature

Hot water 25 35-45 °C 36 – 47 °C 6.04 – 12.08 %

production

District Heating 25 50 - 60 52.6 – 63.66 15.1 – 21.11 %

Power Plant 45 60 - 80 61.57 – 83.67 °C 9.06 – 21.14 %

Organic Rankine 25 °C 32 - 65 °C 32.73 – 69.11 °C 4.3 - 24.16 %

Cycle

Absorption cooling 50 °C 70 - 90 °C 72.1– 94.19 °C 12.1 - 24.16 %

Conclusion

In this report we reviewed data center cooling via water cooling mechanism. We
applied Internal Flow concepts to develop the model and obtain the temperature
profiles and working efficiency. Then we looked at various waste heat recovery
technologies and their individual operating temperatures and efficiency. District
heating seems to be most appropriate technique to realize heat waste recovery
goals while at the same time keeping data center equipment working conditions a
priority.

References

1. A review of data center cooling technology, operating conditions and the

corresponding low-grade waste heat recovery opportunities - ScienceDirect
2. Table 8.1 from Fundamentals of Heat and Mass Transfer for Nusselt number.
3. Eqn 8.45a from Fundamentals of Heat and Mass Transfer for relating outlet, inlet
and surface temperatures in problem.
4. Calculation