COMPACT LIQUID COOLING COMPACT MODULE

专利摘要:
The invention relates to a computer server liquid cooling module, comprising: a pump (101), a fan (117), a heat exchanger (102), at least two ventilation grids (105, 106, 107), a cleared central longitudinal space (109) disposed between the pump (101) and the exchanger (102) to facilitate air flow from a grid (105) from a short side wall (133) to a grate (106, 107) of the other short side wall (134), this air flow being driven by the fan (117), a secondary hydraulic circuit portion (8), for the circulation of a heat transfer fluid, located in the liquid cooling module, comprising no bypass which would allow the pump (101) to operate in a closed circuit and which would be likely to clog up this open longitudinal space (109), an electronic control card (103) arranged in the longitudinal extension of this central longitudinal space released (109) so as to be directly licked by this flow of air.
公开号:FR3057341A1
申请号:FR1659755
申请日:2016-10-10
公开日:2018-04-13
发明作者:Jean-Christophe Bonnin；Elyes Zekri
申请人:Bull SA；
IPC主号:

专利说明:

(57) The invention relates to a liquid cooling module for a computer server, comprising: a pump (101), a fan (117), a heat exchanger (102), at least two ventilation grilles (105, 106, 107 ), a cleared central longitudinal space (109) disposed between the pump (101) and the exchanger (102) so as to facilitate an air flow from a grid (105) from a short side wall (133) towards a grid (106, 107) of the other short side wall (134), this air flow being driven by the fan (117), a portion of secondary hydraulic circuit (8), for the circulation of a fluid coolant, located in the liquid cooling module, comprising no bypass which would allow the pump (101) to operate in a closed circuit and which would be likely to clutter this cleared longitudinal space (109), an electronic control card (103) arranged in the longitudinal extension of this longitudinal space central clear (109) so as to be directly licked by this air flow.
COMPACT LIQUID COOLING MODULE
COMPUTER SERVER
FIELD OF THE INVENTION
The invention relates to the field of liquid cooling modules for computer servers. One or generally several liquid cooling modules cool one or generally several computer servers, often grouped inside a single computer cabinet. The liquid cooling module used is compact, the space inside the computer cabinet being limited.
BACKGROUND OF THE INVENTION
According to a first prior art, it is known to use a relatively contact liquid cooling module. To do this, one or more of the key components of the liquid cooling module, such as the heat transfer liquid circulation pump in the secondary hydraulic circuit, are extracted from the liquid cooling module and are placed outside this cooling module.
The disadvantage of this first prior art is that it only displaces the problem of available space outside the liquid cooling module, or even outside the computer cabinet. Another disadvantage is that the liquid cooling module is no longer autonomous, it becomes more dependent on elements outside it.
According to a second prior art, it is known to use a less compact liquid cooling module. For this, the large key components of the liquid cooling module such as the circulation pump of the heat transfer liquid in the secondary hydraulic circuit, are maintained in the liquid cooling module and are arranged so as to be well ventilated, but the outer casing of this module liquid cooling remains relatively large and therefore bulky. The number of these liquid cooling modules which can be installed in a computer cabinet is therefore reduced; However, this computer cabinet, if it already contains a high density of computer servers, will need more liquid cooling modules to cool it, especially since this computer cabinet will integrate a redundant liquid cooling module to improve the safety and keep all its efficiency during cooling in the event of a breakdown of one of the liquid cooling modules in operation, thus avoiding the shutdown of the cooling system which would be accompanied by the shutdown of the operation of the computer servers of the cabinet computer science.
The disadvantage of this second prior art is that it cannot integrate as many liquid cooling modules as desired to cool a computer cabinet containing a high density of computer servers to be cooled, unless this density is reduced, consequently also reducing the profitability of this computer cabinet, taking up a significant place for more modest computing power.
According to a third prior art, it is known to use a relatively compact liquid cooling module. To do this, the key components of the liquid cooling module such as the circulation pump for the heat transfer liquid in the secondary hydraulic circuit or the heat exchanger between the primary and secondary hydraulic circuits are smaller.
One drawback of this compact, small-component liquid cooling module is that it is less powerful. Even if the number of these liquid cooling modules that can be installed in a computer cabinet can be increased, the total cooling power compared to the volume occupied is not optimized.
SUMMARY OF THE INVENTION
The object of the present invention is to provide a cooling module which at least partially overcomes the aforementioned drawbacks.
More particularly, the invention aims to provide an autonomous liquid cooling module, integrating its own key components such as the circulation pump of the heat-transfer liquid in the secondary hydraulic circuit or as the heat exchanger between primary and secondary hydraulic circuits, which is powerful enough and compact enough to be integrated with sufficient cooling efficiency and reduced volume in a computer cabinet already containing a high density of computer servers thus leaving little space for liquid cooling modules, while also guaranteeing a level of ventilation high enough for its critical component (s) such as its electronic control card, in order to reduce or even avoid the risk of overheating of one of its sensitive components.
For this, the invention proposes a particular arrangement of the most important and voluminous components, between them as well as with respect to the air flow passing inside the outer housing itself of a rather flat geometry, all maintaining a sufficient level of ventilation, in particular conveyed by this air flow, with a simplification of the topology of the secondary hydraulic circuit associated with this new internal arrangement of the liquid cooling module.
The invention therefore proposes a new internal arrangement of the components and a simplification of the layout of the secondary hydraulic circuit to facilitate the passage of the air flow, so as to improve the compromise between the compactness of the module and the efficiency of its ventilation. internal, and its cooling power for computer servers external to it.
To this end, the present invention provides a liquid cooling module for a computer server, comprising an external housing integrating components, characterized in that: the external housing has a length, a width and a thickness such that the length is less than double the width and the thickness is less than half the width, the outer case has four side walls, two said long in the length direction and two said short in the direction of the width, a bottom and a cover , and in that the module comprises, among the integrated components: a pump oriented lengthwise of the outer casing and located along a long side wall, a fan, a heat exchanger oriented in the direction of length of the outer casing and located along the other long side wall, at least two ventilation grids respectively located in the two pa short side kings, a clear central longitudinal space arranged between the pump and the exchanger so as to facilitate an air flow from a grid of a short side wall to a grid of the other short side wall, this flow of air being driven by the fan, a portion of the secondary hydraulic circuit, for the circulation of a heat transfer fluid, located in the liquid cooling module, comprising no bypass which would allow the pump to operate in a closed circuit and which would be capable of cluttering this cleared longitudinal space, an electronic control card arranged in the longitudinal extension of this cleared central longitudinal space so as to be directly licked by this air flow.
Preferably, the heat transfer fluid is a heat transfer liquid, for example glycol water. The primary and secondary hydraulic circuits can also contain heat transfer liquids different from each other, respectively.
According to preferred embodiments, the invention comprises one or more of the following characteristics which can be used separately or in partial combination with one another or in total combination with one another.
Preferably, said electronic card comprises two separable parts which are on the one hand a removable logic part without dismantling the cooling module and on the other hand a connection part fixed to the cooling module without being separately removable from it, to which connection part (of the electronic card) are connected all the connections of the components of the liquid cooling module leading to said electronic card. Thus, the maintenance of the electronic card is facilitated, the greatest number of potential failures being likely to occur on the logic part rather than on the connection part. The maintenance of the critical component that constitutes the electronic card, in a compact liquid cooling module, is usually difficult and will often require a significant disassembly of the cooling module, with the complete opening of the external case requiring for example to remove the cover completely of this outer case, or even having to dismantle other components outside the outer case.
Preferably, the thickness of the outer case is less than a third of the width of the outer case. The outer case is more flat, and takes up less space. Advantageously, to accommodate the key components of sufficient power, the thickness of the outer casing is greater than one sixth of the width of the outer casing, or even greater than one fifth of the width of the outer casing: it is preferably worth about a quarter of the width of the outer casing.
Preferably, the pump oriented lengthwise of the outer casing and located along a long side wall is arranged directly against this long side wall. Thus, the central longitudinal space is better cleared, without reducing the efficiency of the pump.
Preferably, the heat exchanger oriented lengthwise of the outer casing and located along the other long side wall, is arranged very close to this other long side wall without any other element between them than a pipe. . Thus, the central longitudinal space is better cleared, without reducing the efficiency of the exchanger, by the way leaving just enough space for a width of the secondary hydraulic circuit pipe between this exchanger and this other long side wall.
Preferably, said electronic card does not include a protective cover and is directly in contact with all the air flow coming from the cleared central longitudinal space. Thus, the ventilation of the electronic card is improved and an additional gain of space obtained. The simplification of the secondary hydraulic circuit greatly reduced the risk of heat transfer fluid leaking on the electronic card.
Preferably, said electronic card dissipates a calorific power of at least 5W, preferably at most 20W, even more preferably between 7 and 10W. Thus, its need for ventilation is greater, and the internal arrangement of the liquid cooling module according to the invention is all the more advantageous.
The length of the outer casing is of course greater than its width itself, of course greater than its thickness. Preferably, the outer case has a length between 60 and 90cm, a width between 50 and 70cm, a thickness between 10 and 20cm, and preferably has a length between 70 and 80cm, a width between 55 and 65cm, a thickness between 13 and 17cm. This geometry of the outer casing is favorable to a well distributed arrangement of the main components allowing better clearing of a central longitudinal space for the air flow. The outer case has for example a length of 76cm, a width of 59.5cm, a thickness of 15cm.
Preferably, the pump has sufficient power to present a differential pressure of between 2.5 and 3.5 bars at a flow rate of between 50 and 100 liters of heat transfer fluid per minute.
Preferably, the cooling module dissipates a calorific power of at least 50kW, preferably at least 60kW.
Thus, only two liquid cooling modules with an additional redundant module are sufficient to cool a computer cabinet of usual size, containing a good density of computer servers.
Preferably, the pump comprises an air guide channeling the air between, on the one hand, the air inlet ventilation grille in the cooling module and, on the other hand, the inlet of the pump. Thus, this avoids re-injecting air which has become hot by circulating inside the outer casing of the module directly at the inlet of the pump, which would otherwise have the consequence of less good dissipation of the heat produced by the engine. this pump.
Preferably, the cooling module comprises a non-return valve located on the section of secondary hydraulic circuit located between the outlet of the pump and the inlet of the heat exchanger. In the event of a failure of the liquid cooling module pump, this prevents forced circulation of heat transfer liquid in the portion of the secondary hydraulic circuit of this module driven by the pump or pumps of the other liquid cooling modules.
Preferably, the cooling module comprises a valve located on a portion of the primary hydraulic circuit located in the cooling module, having the function of indirectly regulating the temperature of the heat-transfer fluid in the secondary hydraulic circuit at the outlet of the heat exchanger, this valve preferably being a proportional ball valve. It is the main component of the liquid cooling module which manages the level of cooling produced by this module, by regulating the arrival of cool coolant in the primary hydraulic circuit coming from the cold source external to this liquid cooling module.
Preferably, one of the ventilation grilles is a first ventilation grille for air outlet from the cooling module and is located just downstream of said electronic card. Thus, the ventilation of the electronic card is preferred, which is interesting because it is a critical component of the liquid cooling module, which tends to dissipate a lot of heat, especially if a powerful electronic card with multiple functions is chosen.
Preferably, one of the ventilation grids is a second ventilation grille for air outlet from the cooling module and is located just downstream of said valve. Thus, ventilation of the valve which is another component which tends to dissipate a lot of heat, is also preferred.
Preferably, the sum of the areas of the air outlet ventilation grilles is equal to the area of the air inlet ventilation grid. Thus, the air flow is better fluidized, the air flowing through the interior of the outer casing of the module, practically without pressure drop.
Preferably, the heat exchanger is a side-lying exchanger, preferably a plate exchanger, even more preferably a plate and cross-flow exchanger. Thus arranged, the naturally bulky exchanger easily fits into a rather flat outer casing. The type of exchanger chosen optimizes the compromise between its power supplied and its volume occupied.
Preferably, an external insulation layer surrounds on the one hand the exchanger and on the other hand the pipe or pipes of a portion of primary hydraulic circuit located in the cooling module, so as to avoid condensation on their walls. external, even when the temperature of said external walls is lower than the dew temperature of the cooling module. Thus, the risk of condensation water flow is reduced or even avoided, which could have two drawbacks, namely on the one hand damaging another component of the module or at least disturbing its operation, and on the other hand avoiding triggering a false alarm at the level of the leak detector which advantageously contains the liquid cooling module.
Preferably, the cooling module includes a liquid leak detector located in the bottom of the outer case. This leak detector is used to issue an alarm in the event of a heat transfer fluid leak which could damage one or more components of the module or at least disrupt their operation. This leak detector preferably triggers an alarm only in the event of a significant leak, micro-leaks without incidence and without risk on the operation of the liquid cooling module then being advantageously not taken into account and not likely to stop without reason valid operation of the liquid cooling module.
Preferably, the fan is the pump fan and it is coupled to the electric shaft of the pump motor, the motor of which pump is then cooled by air. Thus, this pump fan fulfills two functions simultaneously, on the one hand to cool the pump motor, the pipe part of the pump being cooled by the passage of heat transfer liquid, and on the other hand to drive or facilitate the drive of the air flow in the cleared central longitudinal space located inside the outer case of the liquid cooling module. Alternatively, in the case, for example, of cooling with water or other heat transfer liquid both to the pump motor and to the rest of the pump, the pump then no longer having a fan, the air flow can be facilitated by the ventilator of another component, a ventilator of reduced size added for this purpose, or more simply but less effectively in certain cases by the natural convection of the air between grids of entry and exit.
Other characteristics and advantages of the invention will appear on reading the following description of a preferred embodiment of the invention, given by way of example and with reference to the accompanying drawings.
Another object of the invention is to provide a computer cabinet whose cooling system offers a better compromise, between the space required in the computer cabinet and the robustness in the event of a breakdown of one of the cooling modules in operation, so as to less disrupt the operation of the computer cabinet to avoid degrading its computing performance.
According to this other object of the invention, air cooling is not retained, because it is intrinsically limited by the heat capacity of the air significantly lower than that of a heat transfer liquid. However, with a computer cluster grouping together a large number of computer cabinets with a high density, each computer cabinet grouping together a significant number of computer servers stacked one on top of the other with a high density, the cooling performance requirements seem too high to be satisfied by air cooling.
According to this other object of the invention, water cooling with components deported from the cabinet is not retained, because it requires large pooled elements, such as circulation pumps for heat transfer fluid, which may have two drawbacks. The first drawback lies in the fact that these large elements take up an important place, which can become critical with a computer cluster grouping together a large number of computer cabinets with a high density, each computer cabinet grouping together a significant number of computer servers. stacked on top of each other with high density. The second drawback lies in the fact that these shared elements can prove to be weak links in the entire system, by the very fact of their mutualisation, in the event of a breakdown. Indeed, not only these elements must be stopped, but also all the computer cabinets which depend, for their cooling, on these elements in failure, as well as of course all their computer servers, which can lead to a serious decrease in computing performance for the IT cluster, in the event of failure of certain elements of its IT cooling system.
According to this other object of the invention, water cooling with two 1 + 1 redundant cooling modules raises the following dilemma. In a first option, the redundancy of the modules is preserved, and the cooling system is robust in the event of failure; on the other hand, it becomes necessary to oversize it notably for cooling under normal conditions, since a cooling module of size as large as that which cools the entire computer cabinet remains at rest and therefore unused. In a second alternative option, the redundancy of the modules is not retained, and the cooling system can be dimensioned as precisely as possible for cooling under normal conditions; on the other hand, this cooling system is fragile in the event of failure even of a single cooling module, because a failure then automatically leads to a very significant drop in the cooling capacity which is halved, quickly requiring the complete shutdown of the associated computer cabinet.
According to this other object of the invention, it has been considered to improve this 1 + 1 redundancy by switching to a redundancy of at least 2 + 1, that is to say at least three cooling modules, of which at least two cooling continuously, and the third at rest remains ready to take over in the event of failure of one or the other of the two cooling modules in operation. A master / slave architecture was considered. If this architecture remains robust in the event of a breakdown of a slave module, it becomes fragile in the event of a breakdown of the master module, this again constituting a weak link in the cooling system. In addition, a number of master cooling modules in addition to the slave cooling modules are required in spare inventory.
This other object of the invention then proposes to keep several cooling modules advantageously interchangeable with each other, all cooling except one of them at rest so as to be able to take over in the event of failure of any of the modules cooling in operation. These cooling modules then communicate with each other by a collaborative protocol, without master / slave, which makes it possible to avoid the weak link, while making up for the failure of a cooling module in operation while ensuring its replacement on site, without stopping either the cooling system or the corresponding computer cabinet, while overall taking up less space inside the computer cabinet. According to this other object of the invention, this collaborative protocol is accompanied by greater autonomy for each cooling module, since it will now have to perform certain tasks that it did not previously perform. When the cooling modules are interchangeable with each other, it also reduces the number of types of spare cooling modules to keep.
According to a preferred embodiment of this other object of the invention, it is a question of cooling, at the level of the computer servers of the same computer cabinet, at least 120 kW in 2 + 1 redundancy. Three modules with a power of 60 kW each are sufficient instead of two modules with a power of 120 kW each according to 1 + 1 redundancy. The gain in space occupied in the computer cabinet is significant. There is therefore no master controller so as not to centralize the management of the cooling and avoid making it a critical point. The decentralization adopted by this other object of the invention has led to using a collaborative operating protocol, therefore without master slave, where all the cooling modules communicate at the same level, without hierarchical relationship between them, they are therefore all equal to each other. .
To this end, this other object of the invention provides a computer cabinet, comprising: at least one computer server, at least one liquid cooling module of this server, characterized in that: the cabinet comprises at least 3 cooling modules liquid communicating with each other by a collaborative protocol without master / slave, so as to operate in redundancy N + l with N greater than or equal to 2, so as to be able to carry out a standard exchange of any one of these cooling modules without stopping cooling of the computer cabinet and without stopping the operation of the server located in the computer cabinet, each of these liquid cooling modules including its own system for regulating cooling and detecting breakdowns.
To this end, this other object of the invention also provides a cooling system comprising at least one liquid cooling module of at least one computer server located in a computer cabinet, characterized in that: the cabinet comprises at least 3 liquid cooling modules which are located in the computer cabinet and which communicate with each other by a collaborative protocol without master / slave, so as to operate in redundancy N + l with N greater than or equal to 2, so as to be able to carry out an exchange standard of any of these cooling modules without stopping the cooling of the computer cabinet and without stopping the operation of the server located in the computer cabinet, each of these liquid cooling modules comprising its own system for regulating the cooling and failure detection.
Thus, according to preferred embodiments of this other object of the invention, the liquid cooling module is interchangeable for maintenance without stopping operation of the computer cabinet or of the computer servers it contains. This liquid cooling module then has its own regulation and fault detection electronics, with autonomous and collaborative operation between the liquid cooling modules of the same IT cabinet, without a master controller at the IT cabinet level. Optimized redundancy reduces energy consumption with n modules in operation and only one at rest. The decision to quiesce for the redundant module is taken, on the basis of a collaborative algorithm, by all of the cooling modules of the same computer cabinet.
According to preferred embodiments, this other object of the invention comprises one or more of the following characteristics which can be used separately or in partial combination with one another or in total combination with one or other of all the other objects of the invention.
Preferably, the computer cabinet comprises a secondary hydraulic circuit directly cooling the computer servers, and the liquid cooling modules are connected to this secondary hydraulic circuit in parallel with each other. Thus, the independence of the various cooling modules from each other in the event of a breakdown is better ensured, without requiring bypass circuits which would add to the complexity of the cooling system.
Preferably, the computer cabinet comprises a portion of primary hydraulic circuit intended to be connected to a cold source external to the computer cabinet, and the liquid cooling modules are connected to this portion of primary hydraulic circuit in parallel with one another. . Thus, the independence of the various cooling modules from each other in the event of a breakdown is better ensured, without requiring bypass circuits which would add to the complexity of the cooling system.
Preferably, the computer cabinet comprises a secondary hydraulic circuit directly cooling the computer servers, and the temperature of the heat transfer liquid in this secondary hydraulic circuit is between 20 ° C and 45 ° C. This temperature range ensures proper functioning of the majority of computer servers to be cooled.
Preferably, one of the cooling modules remains off for the majority of the cooling time, preferably for at least 90% of the cooling time. Thus, redundancy is in full swing, and in the event of failure of another of the cooling modules, this one will be ready to take over with significantly less risk of breaking down too.
According to a first alternative, it is always the same cooling module which remains stopped for the majority of the cooling time. In the event of failure of another of the cooling modules, it will be like new and ready to take over without the risk of breaking down too.
According to a second alternative, it is, in turn and periodically, successively each of the cooling modules which remains stopped for the majority of the cooling time. In the event of failure of another cooling module, it will be ready to take over with a relatively low risk of also failing in stride. All the cooling modules in the same computer cabinet will be used regularly and will show a similar degree of wear, which can simplify the management of the cooling module fleet in the same computer room.
Preferably, all of the cooling modules cool together for a minority of the cooling time, preferably only during an initialization phase and / or during a reset phase of the cooling modules and / or during a temporary malfunction of a source. external cold to which the computer cabinet is connected via a portion of the primary hydraulic circuit. Thus, not only the redundant cooling module, that is to say at rest, can take over in the event of a failure of one of the other cooling modules, but in the event of unfavorable conditions when the other cooling modules , without being themselves faulty, no longer manage to provide the desired level of cooling, the redundant cooling module can lend a hand to the other cooling modules which then all continue to operate.
Preferably, the cabinet comprises at least between 3 and 5 liquid cooling modules communicating with each other by a collaborative protocol without master / slave, so as to operate in redundancy N + l with N between 2 and 4, so as to be able to perform a standard exchange of any of these cooling modules without stopping the cooling of the computer cabinet and without stopping the operation of the server located in the computer cabinet. This moderate but sufficient number of cooling modules constitutes a very good compromise between on the one hand the efficiency under normal cooling conditions and on the other hand the robustness in the event of failure.
Preferably, the liquid cooling modules communicate with each other on an equal footing, being subject only to a general manager who, on the one hand manages a fleet of computer cabinets cooling a cluster of computer servers and on the other hand manages for this cluster of computer servers, in addition to its cooling by the fleet of computer cabinets, several other functions among which the start of the cluster of computer servers and their assignment of tasks, for example of calculation tasks. This makes the cooling system more robust as a whole in the event of a breakdown, as it avoids weak links, even at a higher level in the cluster architecture, while improving the autonomy of each of the cooling modules.
Preferably, each cooling module dissipates a calorific power of at least 50kW, preferably at least 60kW. Thus, a simple 2 + 1 redundancy already ensures the efficient dissipation of a large amount of heat while avoiding taking up too much space in the computer cabinet.
Preferably, the computer cabinet comprises a secondary hydraulic circuit directly cooling the computer servers, the computer cabinet comprises a portion of primary hydraulic circuit intended to be connected to a cold source external to the computer cabinet, a heat exchanger cooling the secondary hydraulic circuit via the primary hydraulic circuit, and the temperature differential between the output of the secondary hydraulic circuit and the input of the primary hydraulic circuit is between 0 ° C and 6 ° C. Thus, the heat exchanger presents a good compromise between cooling efficiency and compactness in the cooling module.
Preferably, the cooling regulation and failure detection system of each cooling module comprises one or more temperature sensors, one or more pressure sensors, a water leakage sensor, an opening angular position sensor valve. All of these sensors ensure correct operation of the cooling module under normal cooling conditions, while ensuring significant reactivity in the event of a breakdown or malfunction.
Preferably, the cooling regulation and fault detection system of each cooling module comprises an actuator of the electrical relay responsible for turning a pump on the cooling module on and off, and a valve actuator responsible for check the opening angle of this valve. These various elements contribute to improving the autonomy of the cooling module.
Preferably, all the cooling modules are interchangeable with one another, preferably identical with each other. This reduces the number of types of spare cooling modules to keep.
Preferably, all the liquid cooling modules are located in the lower part of the computer cabinet, below all the computer servers. Thus, in the event of a limited leakage of heat transfer fluid in a cooling module, the leaking liquid does not run the risk of either trickling onto one of the computer servers or disturbing its operation.
Preferably, the heat transfer fluid circulating in the secondary hydraulic circuit has a residual pressure, when it is no longer pumped, which is greater than 0.8 bars, preferably greater than 1.2 bars, even more preferably greater than 2 bars. Thus, the risk of cavitation of the pump is reduced or even eliminated.
Preferably, the computer cabinet comprises a portion of primary hydraulic circuit and a secondary hydraulic circuit located on either side of a heat exchanger, a valve regulating the flow in the portion of primary hydraulic circuit, a temperature sensor. in the secondary hydraulic circuit located at the outlet of the exchanger, said valve being controlled by said temperature sensor, preferably by a proportional / integrator / derivative (PID) type control. This type of control, simple and effective, is suitable and sufficient to properly regulate the cooling of a computer cabinet.
Preferably, each liquid cooling module comprises its own heat exchanger between a portion of the primary hydraulic circuit and a secondary hydraulic circuit, and its own pump for circulating a heat transfer fluid in the secondary hydraulic circuit. Thus, the autonomy of the cooling module is improved.
Preferably, the only maintenance operation of the cooling system which is authorized on the site of the computer cabinet, is the standard exchange of cooling module which consists on the one hand in removing the faulty cooling module and on the other hand, by replacing it with a reserve cooling module without interrupting either the cooling or the functioning of the computer server (s). This facilitated and secure maintenance is made possible by the structure of the computer cabinet and its cooling system, based on autonomous and advantageously interchangeable cooling modules, proposed by this other object of the invention.
Yet another object of the invention aims to provide a computer cabinet whose cooling system offers a better compromise, between the space required in the computer cabinet and the robustness in the event of failure of one of the cooling modules in operation so as to less disrupt the operation of the computer cabinet to avoid degrading its computing performance.
According to this other object of the invention, it is a question of solving the following dilemma. In a first option, the redundancy of the modules is preserved, and the cooling system is robust in the event of failure; on the other hand, it becomes necessary to oversize it notably for cooling under normal conditions, since a cooling module of size as large as that which cools the entire computer cabinet remains at rest and therefore unused. In a second alternative option, the redundancy of the modules is not retained, and the cooling system can be dimensioned as precisely as possible for cooling under normal conditions; on the other hand, this cooling system is fragile in the event of failure even of a single cooling module, because a failure then automatically leads to a very significant drop in the cooling capacity which is halved, quickly requiring the complete shutdown of the associated computer cabinet.
According to this other object of the invention, it has been considered to improve the prior art by passing to a redundancy of at least 2 + 1, that is to say to at least three cooling modules, at least two cooling permanently, and the third at rest ready to take over in the event of a failure of one or the other of the two cooling modules in operation. A master / slave architecture was considered. If this architecture remains robust in the event of a breakdown of a slave module, it remains fragile in the event of a breakdown of the master module, this again constituting a weak link in the cooling system. In addition, a number of master cooling modules in addition to the slave cooling modules are required in spare inventory.
This other object of the invention then proposes to keep several cooling modules advantageously interchangeable with each other, all cooling except one of them at rest so as to be able to take over in the event of failure of any of the modules cooling in operation. These cooling modules then communicate with each other by a collaborative protocol, without master / slave, which makes it possible to avoid the weak link, while allowing the remediation of the failure of a cooling module in operation as well as its replacement on site. , without stopping either the cooling system or the corresponding computer cabinet, while overall taking up less space inside the computer cabinet. According to this other object of the invention, this collaborative protocol is accompanied by greater autonomy for each cooling module, since it will now have to perform certain tasks that it did not previously perform. When the cooling modules are interchangeable with each other, it also reduces the number of types of spare cooling modules to keep.
According to a preferred embodiment of this other object of the invention, it is a question of cooling at least 120 kW in 2 + 1 redundancy. Three modules with a power of 60 kW each are sufficient instead of two modules with a power of 120 kW each that would otherwise be required. The gain in space occupied in the computer cabinet is significant. There is therefore no master controller in order not to centralize the management of the cooling and to make it a critical point. The decentralization adopted by this other object of the invention has led to the use of a collaborative operating protocol, therefore without master slave, where all the cooling modules communicate at the same level, without hierarchical relation between them, they are therefore all equal between them.
This decentralization will lead to greater robustness of the cooling system in the event of a breakdown. However, unlike a very directive master / slave communication, if several cooling modules communicate with each other on an equal footing, via a collaborative protocol, none being the master of another module which would be its slave, a risk of inconsistency or misunderstanding in this non-hierarchical communication between cooling modules can arise, which can lead to premature changes of state for one or the other of the cooling modules, thus risking , under certain conditions, to cause instability in the cooling system of the computer cabinet as a whole. To reduce this risk of instability, a step of checking the consistency of certain data is explicitly planned before certain changes of state, also accompanied by a verification of the stability of this consistency over time, when this consistency has been observed at less punctually.
This additional double precaution of checking consistency and stability of this consistency over time, retains all its flexibility in the collaborative protocol, the essence of its effectiveness, while reducing or even eliminating the risk of divergence or instability of the overall system. cooling at the level of the computer cabinet which would result from an untimely or at least premature change of state in one or more of the cooling modules of the computer cabinet. Among the possibilities of untimely or at least premature change of state in one or more of the cooling modules of the computer cabinet, it is especially the premature transition from a cooling module to the state of cooling module redundant, that is to say at rest, which according to this other object of the invention, the most critical risk of instability of the cooling system. This other object of the invention seeks in fact, for redundancy, that is to say the putting to rest, of a cooling module, to be able to be stable over time and to last a fairly long period of time. without interruption, in order to maintain a better efficiency of this redundancy. If the redundant cooling module, at rest therefore, spends its time oscillating between rest and operation, it is not much better than if it remained in operation all the time.
To this end, this other object of the invention provides a method of communication between several liquid cooling modules of a cooling system of one or more computer servers, characterized in that: the cooling modules communicate with one another so as to operate in redundancy N + l with N greater than or equal to 2, so as to be able to carry out a standard exchange of any one of these cooling modules without stopping the cooling and without stopping the operation of the server or servers, this communication is provided by a collaborative protocol without master / slave, before going from an active mode where it cools to a backup mode where it no longer cools, the redundant cooling module firstly checks that a set of data is consistent between all these cooling modules and on the other hand that this consistency is maintained for a predetermined period e.
According to preferred embodiments, this other object of the invention comprises one or more of the following characteristics which can be used separately or in partial combination with one another or in total combination with one another.
Preferably, the cooling modules communicate with each other over an Ethernet network. This Ethernet network is very well suited to these local communications with the exchange of simple messages between cooling modules within a cabinet.
Preferably, this Ethernet network is also the network on which external commands pass to the computer servers and which is the general network of computer cluster grouping together several computer cabinets which can participate in the execution of the same computer task. This makes the existing Ethernet network profitable and there is no need to increase the complexity of the cooling system by adding an additional dedicated network.
Preferably, during an initialization phase, each cooling module diffuses at least its identifier and an identifier of the computer cabinet in which it is located. Thus, the various cooling modules of the same group located in the same computer cabinet can be sure to quickly touch the other cooling modules of this group, when they do not yet know them.
Preferably, during this initialization phase, each cooling module having received the identifier of another cooling module located in the same computer cabinet as it, sends it a targeted message acknowledging receipt and communicating its own identifier and an identifier of their common computer cabinet, so as to form an information exchange group isolated from the other cooling modules belonging to other information exchange groups. Thus, the various cooling modules of the same group located in the same computer cabinet can set up intra-group communication with all the cooling modules concerned without being interfered with by the cooling modules of other groups.
Preferably, after the completion of this initialization phase, each cooling module periodically sends its data, with a refresh period, its data to the other cooling modules of the information exchange group that it has identified. Thus, all the cooling modules of the same computer cabinet have up-to-date knowledge, possibly almost in real time, of the data of the other cooling modules of the computer cabinet, which improves and simplifies the functioning of the communication between modules, especially since this communication is based on a collaborative protocol for which it is all the more interesting that the cooling modules have updated data as often as possible.
Preferably, each of the cooling modules can have at least the following two statuses: in the active mode, an autonomous status, in which the cooling module cools properly, but fails to synchronize with all the other cooling modules, and still in active mode, a regulated status, in which the cooling module cools properly, and manages to synchronize with all the other cooling modules. Thus, the existence of these two modes of operation improves the overall operation of the cooling system, on the one hand by allowing efficient cooling even if not optimized in the absence of synchronization between the cooling modules then allowing all the same operation of the computer servers of the computer cabinet, and on the other hand by offering optimized cooling in the presence of synchronization between the cooling modules which will allow it to pass a cooling module in redundancy, that is to say at rest, in optimal conditions.
Preferably, each of the cooling modules can have at least the following two statuses: in the event of a fault, a faulty status, in which the cooling module has stopped cooling properly when it should continue to cool properly, in the mode backup, a redundant status, in which the cooling module is at rest but remains ready to immediately replace another cooling module that becomes faulty. The failed status of one of the cooling modules will alert the other cooling module to the redundant status, allowing it to replace it in the sufficient cooling operation of the computer servers in the computer cabinet.
Preferably, when a cooling module goes into faulty status, it stops its heat transfer fluid circulation pump in the secondary hydraulic circuit itself. In the absence of a master cooling module, each cooling module has additional tasks to perform on its own to ensure better running of the overall cooling system.
Preferably, when one of the cooling modules initially succeeds in entering autonomous status, the computer cabinet is started and the computer servers it also blocks. Thus, on the one hand, the functioning of the computer cabinet starts very quickly, while checking on the other hand that a minimum cooling is already available, in order to avoid having to stop in catastrophe a computer cabinet starting to function but without sufficient cooling.
Preferably, the cooling modules each have a set of parameters comprising: a redundant cooling module parameter pointing to the identifier of the cooling module empowered to change to redundant status at the next favorable opportunity, at least one regulation parameter of the cooling pointing to a set value of a parameter regulating cooling. Thus, having at least these few interesting parameters, the cooling modules each have a more complete inventory of the cooling system.
Preferably, the cooling regulation parameter is the target temperature of the heat transfer fluid in the secondary hydraulic circuit at the outlet of the heat exchanger. Indeed, this parameter is specially representative of the proper functioning of the cooling system, better guaranteeing the fact that the temperature of the computer servers does not approach the admissible limit.
Preferably, each cooling module has a consistency indicator which is positive when the following three conditions are simultaneously fulfilled: said cooling module has received from all the other cooling modules the values, of at least the parameter of redundant cooling module and the cooling regulation parameter, updated for less than a first predetermined duration, all the values of the redundant cooling module parameter received are equal to its own value of the redundant cooling module parameter, during more than a second predetermined duration, all the values of the cooling regulation parameter received are equal to its own value of the cooling regulation parameter, for more than a third predetermined duration, which is negative if at least one of these three conditions are not met, said module of r cooling not going from autonomous to regulated status only when its consistency indicator becomes positive. It is this consistency indicator which allows all the cooling modules to check both effectively and simply that their synchronization between them is achieved, and that soon favorable conditions for an optimization of the transition to redundancy of one of between them should come true.
Preferably, the first duration is worth at least double the period of data refresh by the cooling modules, the first duration preferably being between 1 and 10 seconds, even more preferably between 2 and 10 seconds. These durations improve the responsiveness of the cooling system in the event of drift, without significantly increasing the risk of instability of the cooling servo.
Preferably, the second and third durations are between 5 and 60 seconds, even more preferably between 10 and 60 seconds, the second and third durations being advantageously equal to each other. These durations improve the responsiveness of the cooling system in the event of drift, without significantly increasing the risk of instability of the cooling servo.
Preferably, when a cooling module no longer communicates its data to the other cooling modules, its data stored in memory is no longer taken into account in the evaluation of the consistency indicators of the other cooling modules. Thus, the communication between cooling modules and their decision-making associated with the data communicated is no longer polluted by obsolete data no longer corresponding to the actual state of the cooling module that they were supposed to represent.
Preferably, each cooling module has a stability indicator which is positive when the following three conditions are simultaneously met: the consistency indicators of all the cooling modules have been positive for at least a fourth duration, preferably greater than the first, second and third durations, no cooling module has received a cooling malfunction alarm, at most only one of the cooling modules is in the redundant status, all the modules or all the other cooling modules are in the regulated status, which is negative if at least one of these three conditions is not fulfilled, said cooling module only passing from regulated status to redundant status when the following two conditions are fulfilled simultaneously: its redundant cooling module parameter points to its own cooling module identifier, its indicator of stability becomes positive. It is this stability indicator which allows all the cooling modules to check both effectively and simply not only that their synchronization with one another is achieved but also that this synchronization remains stable over time, and that the favorable conditions expected for an optimization of the transition to redundancy of one of them is carried out.
Preferably, the fourth duration is greater than f minute, is preferably between 2 and 5 minutes. These durations improve the responsiveness of the cooling system in the event of drift, without significantly increasing the risk of instability of the cooling servo.
Preferably, if all the cooling modules remain in the autonomous status for at least a fifth predetermined duration, then there is an intervention by an operator external to the computer cabinet, this fifth duration preferably being greater than 10 minutes. Indeed, if the cooling modules have all reached autonomous status, it is because they can operate, while if they fail to synchronize in a reasonable time, it is probably that there is a problem on another level, therefore difficult to solve by cooling modules alone and operator intervention, which we try to minimize because of its cost, then becomes very useful and therefore profitable.
Preferably, when a cooling module has gone into failed status, its data is no longer taken into account in the evaluation of the consistency indicators of the other cooling modules. Thus, the communication between cooling modules and their decision-making associated with the data communicated is no longer polluted by obsolete data no longer corresponding to the actual state of the cooling module that they were supposed to represent.
Preferably, the data sent by each cooling module to the other cooling modules includes: an identifier of its group of cooling modules intended to communicate together and located in the same computer cabinet, cooling together a group of computer servers located in this computer cabinet, its own cooling module identifier, the value of its redundant cooling module parameter, the value of its cooling regulation parameter, a binary parameter corresponding to the presence or absence of a cooling malfunction. Thus, having at least these few interesting parameters and exchanging their values between them, the cooling modules each have an even more complete inventory of the cooling system.
Preferably, the refresh period is between 0.5 and 2 seconds. This duration improves the reactivity of the cooling system in the event of drift, without significantly increasing the risk of instability of the cooling servo.
Preferably, when one of the cooling modules goes into the failed state, then: its redundant cooling module parameter points to its own cooling module identifier, an order is sent to the other cooling modules to point their redundant cooling module on the identifier of this cooling module that has become faulty, the cooling module which is in redundant status, goes into regulated status or into autonomous status, the electrical supply of all the functional elements of this module cooling faulty, except for its electronic control card, is deactivated. Thus, the passage of the indicator between the faulty cooling module and the redundant cooling module before replacing it, is made more fluid.
Preferably, when one of the cooling modules has gone into the failed status, it can then go into the excluded status, and then: a specific command from the operator decides to exclude this cooling module, a specific command of the operator deciding the inclusion of this cooling module will be necessary to restart the excluded cooling module, a simple restart of his electronic control card does not allow said restart.
Preferably, when one of the cooling modules leaves the excluded status, then: a specific command from the operator decides the inclusion of this cooling module, the electrical supply of all the functional elements of this cooling module become faulty is reactivated.
Thus, the decision of exclusion having been taken explicitly, to include again the corresponding cooling module in the cooling system in operation, it is more prudent and more secure, to require an explicit decision also, this in order to reduce the risk instability of the overall cooling system.
Preferably, when one of the cooling modules has gone into autonomous status or into regulated status or into redundant status, it can then go directly to excluded status without going through the faulty status, and then: a specific command of the operator decides to exclude this cooling module, a specific command from the operator deciding the inclusion of this cooling module will be necessary to restart the excluded cooling module, a simple restart of his electronic card control not allowing said restart. In addition to the failure, other types of malfunction that pose a risk to the overall cooling system are likely to result in the exclusion of the cooling module subject to these other types of malfunction.
All the various objects of the invention, as well as all their preferred embodiments, can be combined with one another.
Other characteristics and advantages of the invention and of the other objects of the invention will appear on reading the following description of a preferred embodiment of the invention, given by way of example and with reference to the drawings attached.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 very schematically represents an example of a computer cabinet according to an embodiment of the invention.
FIG. 2 schematically represents an example of a structure of primary and secondary hydraulic circuits in the liquid cooling modules of a computer cabinet according to an embodiment of the invention.
Figures 3 and 4 schematically represent an example of communication between different liquid cooling modules of the same computer cabinet during an initialization phase according to an embodiment of the invention.
FIG. 5 schematically represents an example of an operating diagram of a liquid cooling module communicating with the other liquid cooling modules of the same computer cabinet according to a collaborative communication protocol according to an embodiment of the invention.
FIG. 6 represents the internal arrangement of an example of a compact liquid cooling module according to an embodiment of the invention.
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 very schematically represents an example of a computer cabinet according to an embodiment of the invention.
A computer cabinet 1 contains an Ethernet bus 2 on which one or preferably several computer servers 3 communicate on the one hand and liquid cooling modules 4, 5 and 6 on the other hand. For example, during the operation of the computer servers 3 realizing for example one or more calculations, in cooperation with one another or separately, the cooling modules 4 and 5 are in operation and cool the computer servers 3, while the cooling module 6 is redundant, that is to say at rest but ready to take over and replace immediately, without the intervention of an operator external to the computer cabinet 1, one of the other cooling modules 5 or 6, if it becomes faulty.
The computer server (s) 3 are mounted on a frame of the computer cabinet 1 and are crossed by a secondary hydraulic cooling circuit conveying a heat-transfer fluid or liquid with a high calorific capacity much greater than that of air, this fluid or heat transfer liquid coming from the cooling modules in operation which are modules 4 and 5.
The Ethernet network 2 allows communication between the controllers of each of the cooling modules 4 to 6. Thus, a failure in this network 2 results in a degraded operating mode of this otherwise robust cooling system, in particular in the event of a failure of one of the cooling modules 4 or 5, the redundant cooling module 6 immediately replacing the faulty cooling module, for example the module 4.
FIG. 2 schematically represents an example of a structure of primary and secondary hydraulic circuits in the liquid cooling modules of a computer cabinet according to an embodiment of the invention.
A primary hydraulic circuit 7 supplies the three liquid cooling modules 4 to 6 with cool coolant from a cold source 9 external to the liquid cooling modules 4 to 6. The liquid cooling modules 4 to 6 return to this cold source 9 coolant heated by passing through the liquid cooling modules 4 to 6. The cold source 9 again cools this coolant which begins another round in the primary hydraulic circuit 7. The primary hydraulic circuit 7, coming from the cold source 9, is divided by a divider 71 into three branches of the primary hydraulic circuit 7 which are parallel to each other and respectively pass through the three liquid cooling modules 4 to 6. At the outlet of the three liquid cooling modules 4 to 6, a coupler 72 groups these three branches of primary hydraulic circuit 7 parallel to each other to reconstitute the primary hydraulic circuit 7 going to the cold source 9.
These three liquid cooling modules 4 to 6 in turn supply the computer servers 3 with cooled heat transfer liquid. The computer servers 3 return to the liquid cooling modules 4 to 6 the heat transfer liquid heated by the thermal energy dissipated by these computer servers 3 and evacuated by this heat transfer liquid. The liquid cooling modules 4 to 6 again cool this heat-transfer liquid which begins another turn in the secondary hydraulic circuit 8. The secondary hydraulic circuit 8, coming from the computer servers 3, is divided by a divider 81 into three branches of circuit secondary hydraulic 8 parallel to each other and passing through the three liquid cooling modules 4 to 6 respectively. At the outlet of the three liquid cooling modules 4 to 6, a coupler 82 groups these three branches of secondary hydraulic circuit 8 parallel to each other to reconstitute the secondary hydraulic circuit 8 going to the computer servers 3.
The primary hydraulic circuit 7 and the secondary hydraulic circuit 8 are not in fluid contact, that is to say that the heat transfer liquids of these two hydraulic circuits do not mix with each other. The primary hydraulic circuit 7 and the secondary hydraulic circuit 8 are in thermal contact, that is to say that the heat transfer liquids of these two hydraulic circuits exchange heat with each other, during their passage through the exchangers 40, 50 and 60 liquid cooling modules 4 to 6.
The cooling module 4 includes an exchanger 40, a pump 41, a valve 42, a PID (proportional integrator-derivative) type servo 43, an upstream pressure sensor 44, a downstream pressure sensor 45, an upstream secondary temperature sensor 46, a downstream secondary temperature sensor 47, an upstream primary temperature sensor 48, a downstream primary temperature sensor 49.
The cold heat transfer liquid from a branch of the primary hydraulic circuit 7 enters the liquid cooling module 4, passes through the exchanger 40 where it is heated by heat exchange with the hot heat transfer liquid from a branch of the secondary hydraulic circuit 8, passes through the valve 42, the opening of which regulates its flow rate through the liquid cooling module 4, then emerges from the liquid cooling module 4. The temperature of this cold heat-transfer liquid is measured just after it has entered the liquid cooling module 4 through the upstream primary temperature sensor 48. The temperature of this heated coolant is measured just before it leaves the liquid cooling module 4 by the downstream primary temperature sensor 49. The drive pumping of the coolant into the primary hydraulic circuit 7 is performed by one or more pumps located outside the cooling modules liquid 4 to 6, and possibly shared by the liquid cooling modules of several different computer cabinets. The temperatures measured by the primary temperature sensors 48 and 49 make it possible to verify the correct operation of the primary hydraulic circuit 7.
The hot heat transfer liquid of a branch of the secondary hydraulic circuit 8 enters the liquid cooling module 4, is driven by the pump 41, passes through the exchanger 40 where it is cooled by heat exchange with the cold heat transfer liquid of a branch of the primary hydraulic circuit 7, then out of the liquid cooling module 4.
The temperature of this hot heat transfer liquid is measured just after it enters the liquid cooling module 4 by the upstream secondary temperature sensor 46. The temperature of this heated heat transfer liquid is measured just before it leaves the liquid cooling module 4 through the downstream secondary temperature sensor 47. The temperatures measured by the secondary temperature sensors 46 and 47 make it possible to verify the correct operation of the secondary hydraulic circuit 8. The secondary circuit 8 ensures the circulation of the coolant or heat transfer liquid, in the internal loop in the computer cabinet, at a temperature between 20 ° C and 45 ° C. The temperature sensor 47 is able to measure the temperature at the outlet of the secondary hydraulic circuit 8 from the cooling module 4, the flow of coolant from the primary hydraulic circuit 7 being maintained at a flow rate chosen so that the temperature at the outlet of the secondary hydraulic circuit 8 is equal to a threshold temperature. The temperature at the input of the secondary cooling circuit 8 which passes through the computer servers 3, measured by the upstream secondary temperature sensor 46, is kept constant to optimize the cooling of the electronic components of their calculation blades. The temperature regulation is autonomous for each of the liquid cooling modules 4 to 6, to ensure a constant temperature at the input of the computer servers whatever their heat dissipation.
The primary hydraulic circuit 7 relates to the circuits connected to the hydraulic network of the infrastructure of the client user of the computer cluster grouping together all of the computer cabinets, for example in a computer room, and the secondary hydraulic circuit 8 relates to the hydraulic circuits connected to the computer cabinet cooling. The cooling module 4 comprises a portion of primary hydraulic circuit 7 comprising an outlet capable of being connected to the inlet of a client primary hydraulic network and an inlet capable of being connected to the outlet of the client primary hydraulic network 7. The cooling module 4 also includes a portion of secondary hydraulic circuit 8 comprising an outlet connected to the inlet of the secondary cooling circuit 8 of the computer cabinet and an inlet connected to the outlet of the secondary cooling circuit 8 of the cabinet. computer science.
Furthermore, in the computer cabinet, the inlet and outlet of the primary hydraulic circuit 7 of each cooling module 4 to 6 are provided with quick anti-drip connectors making it possible to easily connect and disconnect the portion of primary hydraulic circuit 7 cooling modules 4 to 6 to the rest of the primary hydraulic network 7 of the computer room of the computer cluster.
The pressure of the heat transfer liquid in the branch of the secondary hydraulic circuit 8 is measured just upstream of the pump 41 by the upstream pressure sensor 44, as well as just downstream of the pump 41 by the downstream pressure sensor 45, in order to control the proper functioning of the pump 41, and in order to be able to stop the pump 41 in the event of a malfunction thereof. The pump 41 has sufficient power to circulate the secondary coolant in the internal loop at a pressure of about 3 bars at a flow rate of about 75 liters per minute of secondary coolant which is for example water glycolée. The pump 41 delivers a constant flow without pressure fluctuations generating vibrations thanks to the shape of the conduits of the pipes of the portion of secondary hydraulic circuit 8 located in the cooling module 4. The pump 41, which is capable of controlling the flow of the heat transfer fluid in the secondary hydraulic circuit 8, is associated with control means able to control this pump 41. The heat transfer liquid in the secondary cooling circuit 8 preferably has a static pressure greater than or equal to 2 bars. Keeping this heat transfer liquid under pressure allows the pump 41 to be protected from any risk of cavitation, and the cooling system of the computer cabinet to function even in the event of a leak.
The inlet and outlet of the secondary hydraulic circuit 8 of each cooling module 4 to 6 are provided with quick anti-drip connectors making it possible to easily connect and disconnect the portion of the secondary hydraulic circuit 8 of the cooling modules 4 to 6 to the rest of the circuit. secondary hydraulic 8 of the computer cabinet cooling the computer servers 3.
In the exchanger 40, 60 kW of heat are exchanged between the secondary hydraulic circuit 8 which gives them and the primary hydraulic circuit 7 which takes them, via a plate exchanger 40, of sufficient size to exchange this power with similar or identical flow rates, or at least of the same order of magnitude, on both sides and an approach temperature of 4 ° C, this approach temperature corresponding to the difference between the external upstream temperature measured by the temperature sensor upstream primary 48 the downstream internal temperature measured by the downstream secondary temperature sensor 47. The heat exchanger 40 is capable of cooling the heat-transfer fluid passing through the secondary hydraulic circuit 8 by dissipation of heat through the primary hydraulic circuit 7.
The servo 43 regulates the opening of the valve 42, and therefore the flow of the heat transfer liquid from the primary hydraulic circuit 7 into the liquid cooling module 4, as a function of the temperature of the heat transfer liquid from the secondary hydraulic circuit 8 at the outlet of the module. liquid cooling 4, measured by the downstream secondary temperature sensor 47.
The liquid cooling module 4 also includes means for controlling the proper functioning of the module 4 and for detecting faults, as well as means for controlling the correct temperature regulation of the heat transfer fluid passing through the secondary hydraulic circuit 8, which are in particular the various temperature and pressure sensors associated with an electronic card shown in FIG. 6 below. The electronic control card for each liquid cooling module is cooled by the fan responsible for cooling the pump motor of this liquid cooling module.
The cooling module 5 includes an exchanger 50, a pump 51, a valve 52, a PID (proportional integrator-derivative) type servo 53, an upstream pressure sensor 54, a downstream pressure sensor 55, an upstream secondary temperature sensor 56, a downstream secondary temperature sensor 57, an upstream primary temperature sensor 58, a downstream primary temperature sensor 59. The cooling module 5 is identical to the cooling module 4. The cooling module 5 operates in the same way as that of the cooling module 4.
The cooling module 6 comprises an exchanger 60, a pump 61, a valve 62, a PID (proportional integrator-derivative) type servo 63, an upstream pressure sensor 64, a downstream pressure sensor 65, an upstream secondary temperature sensor 66, a downstream secondary temperature sensor 67, an upstream primary temperature sensor 68, a downstream primary temperature sensor 69. The cooling module 6 is identical to the cooling module 4. The cooling module 6 operates in the same way as that of the cooling module 4.
The three cooling modules 4 to 6 can operate in basic redundancy, that is to say with three active modules with switching to two active modules during the failure or removal of one of these modules. The mode of operation in particular mode as for example during the initialization of the cooling system is the operation in basic redundancy.
The three cooling modules 4 to 6 can operate in optimized redundancy, that is to say with two active modules 4 and 5 and one module 6 in reserve, ready to start if one of the two active modules 4 or 5 is stopped at following a failure or removal of this module. The normal or cruising mode of operation is optimized redundancy operation.
This operation in basic redundancy and / or in optimized redundancy can of course be generalized with n liquid cooling modules, n then being strictly greater than two.
There is no break in service when one of the n liquid cooling modules is stopped or started. Each of the n liquid cooling modules is removable to allow maintenance without stopping the cooling of the computer servers 3 present in the computer cabinet.
Regulation and fault detection is carried out by monitoring temperature sensors 46 to 49, 56 to 59 or 66 to 69 and pressure sensors 44 and 45, or 54 and 55 or 65 and 66, thanks to an electronic card specific to each cooling module 4, 5 or 6.
The dialogue between the cards of the three liquid cooling modules 4 to 6 is carried out via the TCP protocol (“Transmission Control Protocol / Internet Protocol” in English) or the UDP protocol (“User Datagram Protocol” in English) ). The three liquid cooling modules 4 to 6 also send messages to the SNMP standard (“Simple Network Management Protocol” in English) to a general system for monitoring the cabinet and managing the computer room grouping together several cabinets IT. Between the three liquid cooling modules 4 to 6, there is no master / slave system, since the loss of the master controller would then lead to the loss of the entire cooling system, which the invention seeks to avoid by making it more robust to breakdowns.
The change of the liquid cooling module at rest is scheduled to check its availability as a redundant liquid cooling module, i.e. the emergency liquid cooling module, so that it is operational in the event of an emergency .
A sufficient number of liquid cooling modules are active at the same time permanently, to supply the sufficient quantity of flow of the heat-transfer liquid to the computer servers 3; here the liquid cooling modules 4 and 5 operate while the liquid cooling module 6 is at rest, kept in reserve.
The cooling modules 4 to 6 make it possible to supply the coolant at a maximum constant temperature for each of the calculation blades included in the computer servers 3 mounted on the frame of the computer cabinet.
Two, or n strictly greater than two in the case of generalization, liquid cooling modules 4 and 5 are active, while the other liquid cooling module 6 is active or inactive, depending on the possible redundancy mode chosen which can be a basic redundancy or preferably an optimized redundancy. Thus, if one of the active liquid cooling modules 4 or 5 has a malfunction, the controller of the inactive liquid cooling module 6 is informed by the network 2 and starts its liquid cooling module 6, so that the computer cabinet is always cooled by at least two liquid cooling modules. Thus, even in the event of failure or maintenance of one of the liquid cooling modules, the computer cabinet can continue to operate normally; however in the unlikely event of a second failure before repairing the first failure, consisting essentially of the standard exchange of the faulty liquid cooling module, the cooling will of course deteriorate.
The liquid cooling modules 4 to 6 can be dismantled in the event of maintenance without stopping the cooling of the computer servers 3. The liquid cooling modules 4 to 6 make it possible to regulate the heat transfer liquid in temperature, to ensure a constant temperature at the inlet of the computer servers 3 whatever their heat dissipation.
In addition, the cooling modules 4 to 6 are capable of detecting faults by monitoring pressure and temperature sensors using control means specific to each liquid cooling module. The hydraulic components of the cooling module are optimized, on the one hand, in consumption of pressure drops, in order to minimize the hydraulic energy that the pump 41 must supply and thus optimize its size and its electrical consumption, on the other hand , in terms of size in order to improve the compactness of the liquid cooling module. The cooling system according to a preferred embodiment of the invention makes it possible to dissipate 120 kW for two active liquid cooling modules 4 and 5.
The liquid cooling modules 4 to 6 are identical to each other to be interchangeable and to be replaced by a spare module stored on site in the computer room or nearby. A broken liquid cooling module should be repaired at another site, it is strongly advised not to open the liquid cooling module on the computer room site. Replacement liquid cooling modules are available on site.
The only maintenance operation authorized on site is the removal of the faulty liquid cooling module and its replacement by a spare liquid cooling module in a few minutes, without disruption of service at the computer servers 3 of the computer cabinet.
The operation of a liquid cooling module, and therefore the proper cooling of the liquid cooling system, is not or little affected by the presence or not of the additional redundancy module 6, or else by operation in autonomous mode, which is a degraded operating mode, or not, of the additional liquid cooling module 6.
The effects of connections and disconnections during maintenance are managed by the controllers of each liquid cooling module 4 to 6 to mitigate the transient effects and maintain the good cooling of the computer servers 3 of the computer cabinet.
Advantageously, these three cooling modules 4 to 6 are placed in a lower part of the computer cabinet, the computer servers 3 being placed in an upper part of this same computer cabinet. The lower part is located under the upper part, when the computer cabinet is installed (standing) in operation.
The fact of having the cooling modules 4 to 6 under the computer servers 3 makes it possible to avoid that in the event of a significant leak of the heat transfer liquid in one of the cooling modules 4 to 6, this heat transfer fluid does not flow onto the computer servers 3.
Figures 3 and 4 schematically represent an example of communication between different liquid cooling modules of the same computer cabinet during an initialization phase according to an embodiment of the invention.
In FIG. 3, a first liquid cooling module 4 diffuses data representative of its identity by broadcasting 91 by a UDP broadcasting protocol (“Broadcast UDP” in English). Message 91 includes the identifier of the first cooling module, an identifier of its group of cooling modules included in its cooling cabinet, its Internet address. A second cooling module 5 receiving this broadcast 91 returns to the first liquid cooling module 4 a targeted message 92 which contains on the one hand its own data representative of its own identity and on the other hand an acknowledgment of receipt of the broadcast 91. Message 92 includes an acknowledgment of receipt of message 91, the identifier of the second cooling module, an identifier of its group of cooling modules included in its cooling cabinet (which is therefore the same as that of the first cooling module) , its Internet address (the Internet address of the second cooling module). This is a first identification phase which is carried out prior to the communication and data exchange phase described in Figure 4 below. This first phase of identification is carried out in broadcast mode using a TCP or UDP protocol. The table of Internet addresses of the cooling modules of this group can be built locally at each of these cooling modules, then allowing the process for exchanging information updating their operating state to take place. This diffusion is carried out by each of the cooling modules several times for a period of 2 minutes, since it is an asynchronous communication, to allow the other liquid cooling modules to receive this diffusion. Each of the cooling modules then acts both as a client type data transmitter and as a server type data receiver.
In FIG. 4, the first liquid cooling module 4 sends, to the other liquid cooling modules 5 and 6 of the common computer cabinet, periodically a targeted message 93 containing on the one hand a reminder of its data representative of its identity and an update of certain parameters of its operation. Each of the other liquid cooling modules 5 and 6 does the same thing, that is to say sends, to the other liquid cooling modules of the common computer cabinet, periodically, the same targeted message 93 containing d firstly a reminder of its data representative of its identity and an update of certain parameters of its operation. The period used here is a second. The structure of the data exchanged is as follows and can include, in order: the identifier of their common group, the identifier of the cooling module, one or more state variables, one or more cooling parameters, one or more cooling variables. The cooling module modifies its state variables according to its environment. The status of the IT cabinet is built by the general system of control and supervision of all IT cabinets, that is to say by the general system of control and supervision of the IT cluster. For this, this general control and supervision system can interrogate each cooling module, for example using IPMI ("Intelligent Platform Management Interface") commands periodically, for example every second. Alternatively, a script outside the IT cabinet can periodically interrogate, for example every second, the different cooling modules.
Each of the liquid cooling modules 4 to 6 operates independently and regulates its operation without any of the liquid cooling modules 4 to 6 playing the role of a master directing the other modules like slaves, thanks to the communication protocol collaboration between liquid cooling modules 4 to 6. Each of the liquid cooling modules 4 to 6 can start on their own. Each of the liquid cooling modules 4 to 6 knows the topology of its computer cabinet and therefore of the group of associated liquid cooling modules, as well as its group identifier, as well as its position within this group, all this information being communicated to it. by another computer network when the booting process of the computer cabinet is launched. The cooling module knows its group identifier and its position in this group independently of its Internet address which is communicated to it by the DHCP protocol (“Dynamic Host Configuration Protocol” in English).
FIG. 5 schematically represents an example of an operating diagram of a liquid cooling module communicating with the other liquid cooling modules of the same computer cabinet according to a collaborative communication protocol according to an embodiment of the invention.
The cooling modules communicate with each other by a collaborative protocol, without master or slave, they are led to make their decisions most often by unanimity, sometimes by majority.
The cooling module can take different states or statuses, including an off state 10, a start state 11, a drain state 12, a test state 13, a self-test state 14, a preheating state 15, an autonomous status 16, a status regulated 17, failed status 18, excluded status 19, redundant status 20.
In the off state 10, the cooling module is de-energized.
In start-up state 11, the cooling module is energized and it starts up.
In the emptying state 12, the cooling module performs maintenance allowing in particular to empty the water from the portion of the primary circuit of the cooling module.
In test state 13, the cooling module performs a verification test of its correct operation. Once the verification test is completed at the factory of the cooling module manufacturer, this cooling module will be set to excluded status 19, with default cooling parameters. A new cooling module or a reserve cooling module will in principle be placed in the excluded status 19. After manually inserting a cooling module in the computer cabinet, this cooling module will start in the excluded status 19 and will stay there until further notice. Then, the outside operator will then control the cooling parameters of this cooling module with those of the other cooling modules present in the common IT cabinet because they belong to the same group of cooling modules. It is only after this control phase that this cooling module can be included in this group of cooling modules. While this cooling module remains in excluded status 19, it is the value of its own identifier which will remain assigned to its redundant cooling module parameter.
In self-test state 14, the cooling module itself performs a verification test of its correct operation.
In the preheating state 15, the cooling module performs preheating.
In standalone status 16, the cooling module cools properly but it is not yet synchronized with the other cooling modules. If at least one of the cooling modules succeeds in going into autonomous status 16, then the computer cabinet is powered up as well as all the computer servers it contains.
In regulated status 17, the cooling module cools correctly and is already synchronized with the other cooling modules.
In the faulty status 18, the cooling module no longer works correctly and no longer cools properly, it has broken down. In the faulty status 18, the cooling module will de-energize its pump, in particular in two cases, when the risk of condensation becomes too high, or when the pressure in the secondary circuit becomes too low. Before entering the failed status 18, the cooling module will send an alert which will be received by the general manager supervising the IT cabinets of the IT cluster. When all the cooling modules are in faulty status 18, the general manager supervising the computer cabinet can therefore find that they are all in faulty status 18, that all the pumps are stopped, that the cooling is stopped, that it is therefore necessary to switch off the entire computer cabinet, that is to say all the computer servers it contains. As long as a cooling module remains in faulty status 18 and has not gone into excluded status 19, it can be reset by an operator outside the IT cabinet. If, during this reset, the failure 31 has not returned, this cooling module will first go into the autonomous status 16 then possibly into the regulated status 17. If, during this reset, the failure 31 returns, this cooling module will go, on receipt of an exclusion command 36 from an external operator, to the excluded status 19.
In excluded status 19, the cooling module is explicitly excluded from the group of cooling modules. It will need an explicit include command to be reintegrated into the cooling module group. Without this explicit include command, even a power-on 21 or restart 23 command will not result in reintegration into its group of cooling modules. The restart command 23 can also be carried out from practically all the other states or statuses, except from the extinct state 10.
In the redundant status 20. the cooling module is in redundancy, that is to say that it is at rest, and that it is ready to switch to an autonomous status 16 to replace another cooling module of their common group, cool in its place, if this other cooling module has either gone into a faulty status 18 or has been put into an excluded status 19.
The cooling module goes from one state or status to another, either through an command from an outside operator, or through an operation that it performs itself when the corresponding condition is fulfilled.
Among these commands or these operations are a power supply command 21, a drain command 22, a restart command 23, a self-test start command 24, a self-test output command 25, a start command for test 26, a cooling supply command 27, a start failure exclusion command 28, an inclusion command 29, a preheating termination operation 30, a failure 31, a preheating failure 32, an operation for finding data consistency 33, for finding data inconsistency 34, for switching to redundancy 35, for an exclusion command 36.
The power supply control 21 energizes the cooling module and initiates the start of this cooling module. The power supply control 21 switches the cooling module from an off state 10 to a start state 11.
The drain command 22 gives the order to carry out maintenance making it possible in particular to empty the water from the portion of the primary circuit of the cooling module. The drain command 22 switches the cooling module from a start state 11 to a drain state 12.
The restart command 23 instructs the cooling module to perform a restart. The restart command 23 switches the cooling module from a drain state 12 to a start state 11, or from a test state 13 to a start state 11.
The self-test start command 24 instructs the cooling module to start a self-test. The self-test start command 24 switches the cooling module from a test state 13 to a self-test state 14.
The self-test output command 25 instructs the cooling module to stop the self-test. The self-test output command 25 switches the cooling module from a self-test state 14 to a test state 13.
The test start command 26 instructs the cooling module to start a test. The test start command switches the cooling module from a test state 13 to a start state 11.
The cooling supply control 27 instructs the cooling module to start preheating. The cooling supply control 27 switches the cooling module from a start state 11 to a preheat state 15.
The boot failure exclusion command 28 excludes the cooling module from its group of cooling modules after it fails to start. The start failure exclusion command 28 switches the cooling module from a start state 11 to an excluded state 19.
The include command 29 includes or re-includes the cooling module in its group of cooling modules. The inclusion command 29 switches the cooling module from an excluded status 19 to a start state IL When the inclusion command 29 is sent to one of the cooling modules, then the inclusion process begins when this cooling module receives this inclusion command 29, preferably in the form of an IPMI command. This cooling module then assigns the value of its own identifier to its redundant cooling module parameter, which in fact should already be the case. This cooling module goes to start-up state 11, the cooling parameters having normally been previously initialized directly by the operator during the previous exclusion process. If a cooling power supply command 27 is received by this cooling module, it will go to autonomous status 16 first, and to regulated status 17 then if a data consistency observation operation 33 is carried out. Then, if the stability indicator also becomes positive, then this cooling module will go to redundant status 20 by a switchover operation to redundancy 35.
The preheating termination operation 30 stops the preheating of the cooling module. The preheating termination operation 30 switches the cooling module from a preheating state 15 to an autonomous status 16.
Failure 31 is a failure of the cooling module. Failure 31 switches the cooling module to a failed status 18 from an autonomous status 16, or from a regulated status 17, or from a redundant status 20. When the failure 31 occurs for one of the cooling modules, then this cooling module assigns the value of its own identifier to its redundant cooling module parameter, sends a request to the other cooling modules asking them to set their redundant cooling module parameter to the same value as him. Then, the redundant cooling module, leaves the redundant status 20 to pass to the regulated status 17 or rather quickly to the autonomous status 16, because the stability indicator will become negative since one of the cooling modules, the one which has just passed in defaulted status 18, is no longer in regulated status 17.
The preheating failure 32 is a failure of the cooling module occurring during its preheating. The preheating failure 32 changes the cooling module to a faulty status 18 from a preheating state 15.
The data consistency finding operation 33 makes the cooling module consistency indicators positive or leaves positive. The data consistency observation operation 33 changes the cooling module from an autonomous status 16 to a regulated status 17.
The data inconsistency finding operation 34 makes at least one of the consistency indicators of the cooling modules negative or leaves negative. The data inconsistency finding operation 34 switches the cooling module to an autonomous status 16 from a regulated status 17, or from an autonomous status 20. When one of the cooling modules assets goes into faulty status 18, it stops its pump, and the cooling module which was in redundant status 20 starts its pump by itself to go into autonomous status 16.
The switchover to redundancy operation 35 puts a single cooling module at rest with the ability to start cooling immediately in an emergency, in particular to replace another cooling module that has become faulty. The operation for switching to redundancy 35 switches the cooling module from a regulated status 17 to a redundant status 20.
Exclude command 36 excludes the cooling module from its group of cooling modules. The exclusion command 36 switches the cooling module to an excluded status 19 from an autonomous status 16, or from a regulated status 17, or from a faulty status 18, or else from a redundant status 20.
When the exclusion command 36 is sent to one of the cooling modules, then the exclusion process begins when this cooling module receives this exclusion command 36, preferably in the form of an IPMI command. This cooling module then assigns the value of its own identifier to its redundant cooling module parameter, sends a request to the other cooling modules asking them to set their redundant cooling module parameter to the same value as it. Then, the redundant cooling module leaves the redundant status 20 to go either to the autonomous status 16 or to the regulated status 17.
The exclusion command 36 is, for the cooling module to which it is sent, the first step of the process tending to remove a cooling module with a faulty status 18, outside the computer cabinet, in order to carry out maintenance on this cooling module. cooling.
In normal operating conditions, the cooling modules communicate with each other periodically, via a collaborative protocol, in order to exchange a set of data with each other, in order to be able to check first if this set of data is consistent, which will result in a coherence indicator becoming positive, and then if this dataset remains coherently stable over time, which translate into a stability indicator becoming positive.
During this exchange of data, each cooling module controls several things, namely that it controls a possible loss of communication with one or other of the other cooling modules, that it also controls the consistency of the data exchanged. , as well as the stability over time of the consistency of these exchanged data. These checks can be carried out while this cooling module is either in autonomous status 16, or in regulated status 17, or in redundant status 20.
The control of communication loss between cooling modules consists for each of the cooling modules to verify that there have been no problems of communication loss on the Ethernet network on which these cooling modules communicate with each other. Several scenarios can be envisaged on this subject.
First, one of the cooling modules can be missing, that is to say, no longer send on the Ethernet network bus, in which case its data is no longer taken into account in all the data of the group and are no longer selected to perform the data consistency observation operation 33.
Then, two of the three cooling modules can be missing, that is to say, no longer transmit on the Ethernet network bus, in which case only the data of the last cooling module still capable of communicating, therefore to transmit on the Ethernet buses are taken into account in all of the group's data and are used to perform the data consistency observation operation 33.
Finally, one of the cooling modules can be placed in the failed status 18 or in the excluded status 19, in which case its data are no longer taken into account in all of the group's data and are no longer retained to perform the data consistency operation 33.
The consistency check of all the data exchanged can relate to all or part of the data exchanged. Periodically, for example every second, a process will compare if the cooling parameters are equal between all the cooling modules, and if this equality is maintained for at least 10 seconds.
The cooling parameters comprise, on the one hand, the redundant cooling module parameter whose value is that of the identifier of the cooling module which is supposed to switch to redundancy, that is to say at rest, when the conditions will be favorable, that is to say when the switching operation to redundancy 35 will take place, and on the other hand the parameter for regulating the cooling, the value of which is that of the target temperature for regulating the coolant circulating in the secondary hydraulic circuit at the exchanger outlet.
At the level of a global initialization phase of a set of computer cabinets, the general surveillance system of the computer room or an external operator (a human being responsible for the supervision or maintenance of this computer room) , can initialize the cooling parameters of each computer cabinet by sending IPMI commands to all the cooling modules of the corresponding computer cabinet. Normally, these commands will be sent and received by all the cooling modules at almost the same time. As an additional security measure, a safety margin can be taken and the time slot for making these sendings can be extended to 10 seconds. Once each cooling module has received the IPMI commands, it changes its cooling parameters and goes into regulated status 17. The other cooling modules in turn have a time slot of 10 seconds also to receive these parameter values from cooling, update their cooling parameters from these received values, and send their own cooling parameters back to the other cooling modules. If this exchange of data is not carried out within the time allowed, then each cooling module having noted the failure will go into autonomous status 16, since the consistency indicator will be negative.
The consistency indicator remains or becomes positive when the following three conditions are met simultaneously. The first condition is fulfilled when the data set exchanged between cooling modules has been updated for at most 2 seconds, i.e. at most one update sent has been missed by one either of the cooling modules. The second condition is fulfilled when the parameters of the redundant cooling module are equal to each other for all the cooling modules of the group, and this for at least 10 seconds. The third condition is fulfilled when the cooling control parameters are equal to each other for all the cooling modules in the group, and this for at least 10 seconds.
Conversely, the consistency indicator remains or becomes negative when at least one of the three preceding conditions is not or is no longer fulfilled. When the consistency indicator is positive, the cooling module can go from autonomous status 16 to regulated status 17. When the consistency indicator is negative, the cooling module which is in autonomous status 16 remains there. In the event of persistence of a negative consistency indicator, a corrective action is implemented, for example by the external operator, to analyze and correct the one or those of the cooling parameters which are responsible for the persistence of an indicator of negative consistency.
The stability control of all the data exchanged can relate to all or part of the data exchanged. The stability control relates to the same data as that which is the subject of the consistency control. Periodically, for example every second, a process will check that the consistency indicator remains positive for a certain time during which the cooling modules remain synchronized with each other and none of them has received an alarm from cooling malfunction. When the stability indicator remains or becomes positive, the cooling module that is supposed to go to rest, that is to say become redundant, will effectively become or remain so if it is already there.
The stability indicator remains or becomes positive when the following three conditions are met simultaneously. The first condition is fulfilled when all the consistency indicators at the level of the cooling modules remain positive for at least 3 minutes. The second condition is fulfilled when no cooling malfunction alarm has been received by a cooling module. Such a cooling malfunction alarm can for example be received because of a pump or exchanger that has become defective. The third condition is fulfilled when at least two of the three (or n of the n + l) group cooling modules are in regulated status 17, the third (or last) group cooling module being in the regulated status 17, i.e. in redundant status 20.
Conversely, the stability indicator remains or becomes negative when at least one of the three preceding conditions is not or is no longer fulfilled. When the stability indicator is positive, and one of the cooling modules points its cooling module parameter to its own identifier, then this cooling module can go from regulated status 17 to redundant status 20, and only in this case there. When the stability indicator is negative, no cooling module can go into redundant status 20, but must instead remain in regulated status 17.
FIG. 6 represents the internal arrangement of an example of a compact liquid cooling module according to an embodiment of the invention. The cooling module dissipates around 60kW. The cooling module comprises an external housing 100 integrating several components. The heat transfer fluid is a heat transfer liquid, for example brine. A primary hydraulic circuit 7, containing a cross heat transfer liquid, without mixing of heat transfer liquid, but with heat exchange between the heat transfer liquids, a secondary hydraulic circuit 8, containing a heat transfer liquid, inside the outer housing 100 of the module liquid cooling, inside a heat exchanger 102.
The outer housing 100 includes two long side walls 131 and 132 opposite one another, two short side walls 133 and 134 opposite one another, a bottom 135 opposite a cover 136 which is not visible in Figure 6 because it has been removed to show the interior of the outer housing 100. The terms "long" and "short" only mean that the long side walls 131 and 132 are longer than the short side walls 133 and 134. The general shape of the outer case 100 is parallelepiped. The long side walls 131 and 132 are substantially the same length between them. The short side walls 133 and 134 are substantially the same length therebetween. The outer box 100 has a length L of 76cm, a width 1 of 59.5cm and a thickness e of 15cm.
Among the components integrated in the outer housing 100 are a pump 101, a heat exchanger 102, an electronic card 103, a valve 104, an inlet grille 105, two outlet grilles 106 and 107, a check valve. return 108, a cleared longitudinal center space 109, a leak detector 110, pressure sensors 111, a power supply relay 112, temperature sensors 113, an air guide 114, a fan 117, pipes 121 at 123 in the secondary hydraulic circuit 8, pipes 124 and 125 in the primary hydraulic circuit 7. The terms pipes and pipes are used interchangeably.
The pump 101 is oriented in the lengthwise direction of the outer casing 100 and located along the long side wall 132, it is disposed directly against this long side wall 132. The axis of the pump 101 is aligned with the upstream pipe 121 to have a regular supply of the pump 101. This upstream pipe 121 is a flexible pipe for connecting the pump 101, which allows it to compensate for misalignments and allows the dismantling of the pump 101 for maintenance.
The heat exchanger 102 is oriented lengthwise of the outer casing 100 and located along the other long side wall 131, it is arranged very close to this other long side wall 131 without any other element between them that 'a pipeline. The heat exchanger 102 is an exchanger 102 lying on the side to minimize the space requirement, a plate and cross flow exchanger 102, made of copper brazed stainless steel, in order to have a better heat exchange performance.
An external insulation layer surrounds on the one hand the exchanger 102 and on the other hand the pipe or pipes 124 and 125 with a portion of primary hydraulic circuit 7 located in the cooling module, so as to avoid condensation on their external walls, even when the temperature of these external walls is lower than the dew point of the cooling module. This insulation, on the exchanger 102 and on the pipes 124 and 125 of the primary circuit 7, thus avoids a condensation monitoring system capable of triggering an alarm of cooling malfunction without valid reason. The pipes 124 and 125 have elbows at the outlet and at the inlet of the exchanger 102 to minimize the bulk.
The electronic card 103 itself comprises two parts which are on the one hand a logic part 115 and on the other hand a connection part 116. This electronic control card 103 is arranged in the longitudinal extension of this cleared central longitudinal space 109 of so as to be directly licked by the air flow passing through it. The two parts 115 and 116 which can be separated from each other, of this electronic card 103, are on the one hand a logic part 115 which can be dismantled without dismantling the rest of the cooling module and on the other hand a connection part 116 fixed to the bottom 135 of the module without being separately removable. To this connection part 116 are connected all the connections of the components of the liquid cooling module leading to this electronic card 103. This electronic card 103 does not include a protective cover and is directly in contact with all the air flow coming from the central longitudinal space cleared 109. This electronic card 103 dissipates a calorific power of approximately 10W. This electronic card 103 is a centralized electronic regulation card, cooled by the air flow created by the fan 117 of the pump 101.
The valve 104 is located on a portion of the primary hydraulic circuit 7 located in the cooling module, having the function of indirectly regulating the temperature of the coolant in the secondary hydraulic circuit 8 at the outlet of the heat exchanger 102, this valve 104 being advantageously a valve 104 with proportional ball valve. This valve 104 with proportional ball valve, includes a device for adjusting the passage section promoting the linearity of the flow response as a function of the opening angle, with a servomotor mounted lying to minimize the space requirement, with a command electric allowing a precise opening and a rereading of this opening more precise than with a proportional solenoid valve.
The inlet ventilation grid 105 is located in the short side wall 133. The two outlet ventilation grids 106 and 107 are respectively located in the short side wall 134. The grid 106 is a first outlet ventilation grid 106 d air outside the cooling module and is located just downstream from the electronic card 103. The grid 107 is a second grid 107 for ventilating the air outlet from the cooling module and is located just downstream from the valve 104. The sum of the areas of the air outlet ventilation grilles 106 and 107 is equal to the area of the air inlet ventilation grid 105.
The non-return valve 108 is located on the section of secondary hydraulic circuit 8 located between the outlet of the pump 101 and the inlet of the heat exchanger 102. The non-return valve 108 is located near the heat exchanger 102 to create a removable block, which is advantageous to avoid the risk that the flow coming from the other cooling modules enters it when its pump 101 is stopped. .
The cleared longitudinal center space 109 is disposed between the pump 101 and the exchanger 102 so as to facilitate an air flow from a grid 105 from a short side wall 133 to the two grids 106 and 107 of the another short side wall 134, this air flow being driven by the fan 117 of the pump 101.
A portion of secondary hydraulic circuit 8, for the circulation of a coolant, located in the liquid cooling module, does not include any bypass which would allow the pump 101 to operate in a closed circuit and which would be likely to clutter this longitudinal space central cleared 109. This central longitudinal space cleared 109 between the components of the cooling module is sufficient to promote the passage of cooling air between on the one hand the upstream grid 105 and on the other hand the downstream grids 106 and 107, so that both the electronic card 103 and the valve 104 are well cooled.
The leak detector 110 is located in the bottom 135 of the outer casing 100. This leak detector 110 is placed near the drain pipe 125, the bottom 135 of the outer casing 100 is waterproof and can accommodate a significant amount of liquid. in the event of a leak. The leak is only detected if it is relatively large, small leaks are deliberately not taken into account, because they are not very troublesome.
The pressure sensors 111 are located respectively at the inlet and at the outlet of the pump 101 to verify the operation of the latter.
The relay 112 for supplying power to the pump 101 is housed in a sealed housing which accommodates high voltages for supplying the pump 101 with alternating current.
The temperature sensors 113 monitor and regulate the circulation of coolant in the primary 7 and secondary 8 circuits: there is one at each line 121, 122, 124, 125 near their crossing of the wall short side 134.
The air guide 114 channels the air between on the one hand the air intake ventilation grille 105 in the cooling module and on the other hand the inlet of the pump 101. The pump 101 being placed near of the grille 105 situated on the front face of the external housing 100, its fan 117 sucks in fresh air coming from the outside through this grille 105 with an air guide 114 to avoid re-sucking in hot air which had already passed through the cooling module.
The air cooling fan 117 of the pump 101 is coupled to the electric shaft of the pump motor 101.
The pipe 121 located upstream of the pump 101 in the secondary hydraulic circuit 8 and the pipe 122 located between the pump 101 and the exchanger 102 in the secondary hydraulic circuit 8, as well as the pipe 123 located downstream of the exchanger 102 in the secondary hydraulic circuit 8, form the portion of the secondary circuit 8 located in the cooling module.
The line 124 located upstream of the exchanger 102 in the primary hydraulic circuit 7 and the line 125 located downstream of the exchanger 102 in the primary hydraulic circuit 7, form the portion of primary circuit 7 located in the cooling module.
The elbows of the various pipes 121 to 125 are elbows with a large radius of curvature to minimize the pressure drop and not to disturb the flow of heat transfer liquid in these same pipes 121 to 125. The crossings of the short side wall 134 of the housing outside
100 are compact and do not exhibit any significant change in section in order to minimize the pressure drop.
The primary 7 and secondary 8 circuits include flexible connectors which are well aligned with the rigid pipes 121 to 125 to minimize changes in passage cross-section and disturbances in the flow of coolant. Likewise, pressure fluctuations are reduced as well as the vibrations generated in the cooling module and throughout the rest of the computer cabinet. In addition, erosion is also minimized by the regularity of the flow of coolant in the rigid pipes 121 to 125.
Of course, the present invention is not limited to the examples and to the embodiment described and shown, but it is susceptible of numerous variants accessible to those skilled in the art.

权利要求:
Claims (20)
[1" id="c-fr-0001]
1. Liquid cooling module for a computer server, comprising an external box (100) integrating components, characterized in that:
> the outer casing (100) has a length (L), a width (1) and a thickness (e) such that the length (L) is less than twice the width (1) and the thickness (e) is less than half the width (1),> the outer casing (100) has four side walls, two called long (131, 132) lengthwise and two so-called short (133, 134) width, a bottom (135) and a cover (136), and in that the module comprises, among the integrated components:
> a pump (101) oriented lengthwise of the outer casing (100) and located along a long side wall (132),> a fan (117),> a heat exchanger (102) oriented in the direction of the length of the outer casing (100) and located along the other long side wall (131),> at least two ventilation grilles (105, 106, 107) respectively located in the two short side walls (133 , 134),> a clear central longitudinal space (109) disposed between the pump (101) and the exchanger (102) so as to facilitate an air flow from a grid (105) of a short side wall (133) to a grid (106, 107) on the other short side wall (134), this air flow being driven by the fan (117),> a portion of secondary hydraulic circuit (8), for circulation a heat transfer fluid, located in the liquid cooling module, comprising no bypass which allows relating to the pump (101) operating in a closed circuit and which could clutter this cleared longitudinal space (109),> an electronic control card (103) arranged in the longitudinal extension of this cleared longitudinal central space (109) so as to be directly licked by this air flow.
[2" id="c-fr-0002]
2. Cooling module according to claim 1, characterized in that:
> said electronic card (103) comprises two separable parts which are on the one hand a logic part (115) removable without disassembly of the cooling module and on the other hand a connection part (116) fixed on the cooling module without being separately removable, to which connection part (116) are connected all the connections of the components of the liquid cooling module leading to said electronic card (103).
[3" id="c-fr-0003]
3. Cooling module according to any one of the preceding claims, characterized in that:
> the thickness (e) of the outer case (100) is less than a third of the width (1) of the outer case (100).
[4" id="c-fr-0004]
4. Cooling module according to any one of the preceding claims, characterized in that:
> the pump (101) oriented lengthwise of the outer casing (100) and located along a long side wall (132) is arranged against this long side wall (132).
[5" id="c-fr-0005]
5. Cooling module according to any one of the preceding claims, characterized in that:
the heat exchanger (102) oriented lengthwise of the outer casing (100) and located along the other long side wall (131), is arranged very close to this other long side wall (131) without any other element between them than a pipe (123).
[6" id="c-fr-0006]
6. Cooling module according to any one of the preceding claims, characterized in that:
> said electronic card (103) does not include a protective cover and is directly in contact with all the air flow coming from the cleared central longitudinal space (109).
[7" id="c-fr-0007]
7. Cooling module according to any one of the preceding claims, characterized in that:
> said electronic card (103) dissipates a calorific power of at least 5W, preferably at most 20W, even more preferably between 7 and 10W.
[8" id="c-fr-0008]
8. Cooling module according to any one of the preceding claims, characterized in that:
> the outer casing (100) has a length (L) between 60 and 90cm, a width (1) between 50 and 70cm, a thickness (e) between 10 and 20cm, and preferably has a length (L) between 70 and 80cm, a width (1) between 55 and 65cm, a thickness (e) between 13 and 17cm.
[9" id="c-fr-0009]
9. Cooling module according to any one of the preceding claims, characterized in that:
> the pump (101) has sufficient power to present a differential pressure of between 2.5 and 3.5 bars at a flow rate of between 50 and 100 liters of heat transfer fluid per minute.
[10" id="c-fr-0010]
10. Cooling module according to any one of the preceding claims, characterized in that:
> the cooling module dissipates a calorific power of at least 50kW, preferably at least 60kW.
[11" id="c-fr-0011]
11. Cooling module according to any one of the preceding claims, characterized in that:
the pump (101) comprises an air guide (114) channeling the air between on the one hand the air inlet grille (105) of air in the cooling module and on the other hand the inlet of the pump (101).
[12" id="c-fr-0012]
12. Cooling module according to any one of the preceding claims, characterized in that:
the cooling module comprises a non-return valve (108) located on the section (122) of the secondary hydraulic circuit (8) located between the outlet of the pump (101) and the inlet of the heat exchanger (102 ).
[13" id="c-fr-0013]
13. Cooling module according to any one of the preceding claims, characterized in that:
the cooling module comprises a valve (104) located on a portion (124) of the primary hydraulic circuit (7) located in the cooling module, having the function of indirectly regulating the temperature of the heat-transfer fluid in the secondary hydraulic circuit (8 ) at the heat exchanger outlet (102), this valve (104) preferably being a proportional ball valve.
[14" id="c-fr-0014]
14. Cooling module according to any one of the preceding claims, characterized in that:
> one of the ventilation grilles is a first outlet ventilation grille (107) for air from the cooling module and is located just downstream of said electronic card (103).
[15" id="c-fr-0015]
15. Cooling module according to any one of the preceding claims, characterized in that:
> one of the ventilation grilles is a second outlet ventilation grille (106) for air from the cooling module and is located just downstream of said valve (104).
[16" id="c-fr-0016]
16. Cooling module according to claims 14 and 15, characterized in that:
> the sum of the surfaces of the air outlet ventilation grilles (106, 107) is equal to the area of the air intake ventilation grille (105).
[17" id="c-fr-0017]
17. Cooling module according to any one of the preceding claims, characterized in that:
> the heat exchanger (102) is a side-lying exchanger, preferably a plate exchanger, even more preferably a plate and cross-flow exchanger.
[18" id="c-fr-0018]
18. Cooling module according to any one of the preceding claims, characterized in that:
> an external insulation layer surrounds on the one hand the exchanger (102) and on the other hand the pipe or pipes (124, 125) with a portion of primary hydraulic circuit (7) located in the cooling module, so as to avoid condensation on their external walls, even when the temperature of said external walls is lower than the dew point of the module
5 cooling.
[19" id="c-fr-0019]
19. Cooling module according to any one of the preceding claims, characterized in that:
> the cooling module includes a leak detector (110)
10 of liquid located in the bottom (135) of the outer case (100).
[20" id="c-fr-0020]
20. Cooling module according to any one of the preceding claims, characterized in that:
> the fan (117) is the pump fan (101) and is
15 coupled to the electric shaft of the pump motor (101).
1/5

类似技术:

---

公开号 | 公开日 | 专利标题

---

FR3057341A1|2018-04-13|COMPACT LIQUID COOLING COMPACT MODULE

---

CA3040049A1|2018-04-19|Computer cabinet with liquid cooling modules

---

EP3524044B1|2021-05-19|Protocol for communication between a plurality of modules for the liquid cooling of computer servers

---

FR2980060A1|2013-03-15|OPTIMIZATION OF A DOMOTIC SYSTEM UTILIZING A LOCAL NETWORK OF LINE CARRIER CURRENT

---

US20150355630A1|2015-12-10|Unified control of an electronic control system and a facility control system

---

BE1019118A5|2012-03-06|UNIT, METHOD AND SYSTEM FOR TREATING AIR.

---

EP3044640A1|2016-07-20|Automation system

---

EP2065662B1|2020-01-22|Secured water and/or air dispenser for a snow cover system, equipped with a valve controlled by an electrical actuator

---

US9603282B2|2017-03-21|Datacenter and cooling control fault-tolerance using compute resources

---

EP2524871B1|2017-08-23|Autonomous electric power generation system for aircraft and related control method

---

FR2894563A1|2007-06-15|Heat exchange circuit for air conditioning and cooling electronic equipment in aircraft, comprises heat exchangers, first and second heat transfer fluid loops and auxiliary tubes connecting the loops

---

EP3553377B1|2021-10-13|Safety device for steam generator

---

EP3848659A1|2021-07-14|Partitioned heat exchanger, thermal energy recovery unit and associated sterilisation device

---

FR3077690A1|2019-08-09|REGULATING THE OPERATING FREQUENCY OF A TRANSPORT NETWORK AND / OR DISTRIBUTION OF ELECTRICAL ENERGY

---

CA3100919A1|2019-11-28|Heat recuperation system fdr heatsng e.g, greenhouses

---

EP3032180B1|2017-03-01|Thermal device and method for using such a device

---

KR20170081864A|2017-07-13|Method for driving a feed water system of a boiler of a plant by using Run-to-Min Priority, and a computer program therefor

---

FR3082598A1|2019-12-20|MOBILE CASH BACK INSTALLATION

---

EP1858136A1|2007-11-21|Method and device for stopping an electric appliance supplied by a plurality of energy sources, appliance equipped with such a device and system comprising such an appliance

---

EP3942233A1|2022-01-26|Control for supervising communication between a central server and a local control of thermal device being part in a heat or cool distributing network

同族专利:

---

公开号 | 公开日

---

JP2019531556A|2019-10-31|

---

US20200053916A1|2020-02-13|

---

FR3057341B1|2019-05-24|

---

RU2019113938A|2020-11-13|

---

US11083111B2|2021-08-03|

---

BR112019007125A2|2019-07-02|

---

CN110140437A|2019-08-16|

---

EP3524045A1|2019-08-14|

---

CA3040050A1|2018-04-19|

---

WO2018069635A1|2018-04-19|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

EP1448040A2|2003-02-14|2004-08-18|Hitachi, Ltd.|Liquid cooling system for a rack-mount server system|

---

US20050133214A1|2003-12-19|2005-06-23|Teradyne, Inc.|Modular rackmount chiller|

---

US20050247433A1|2004-05-04|2005-11-10|International Business Machine Corporation|Method and apparatus for cooling electronic components|

---

US20110308783A1|2010-06-17|2011-12-22|Mark Randal Nicewonger|Fluid-powered heat exchanger apparatus for cooling electronic equipment|

---

EP2773173A1|2013-02-28|2014-09-03|Bull Sas|Liquid cooling system for computer cabinet|

---

US4493010A|1982-11-05|1985-01-08|Lockheed Corporation|Electronic packaging module utilizing phase-change conductive cooling|

---

US5323847A|1990-08-01|1994-06-28|Hitachi, Ltd.|Electronic apparatus and method of cooling the same|

---

US5772500A|1996-12-20|1998-06-30|Symbios, Inc.|Compact ventilation unit for electronic apparatus|

---

JP2000049479A|1998-07-28|2000-02-18|Fujitsu Ltd|Electronic device|

---

DK2032907T3|2006-06-01|2018-07-02|Google Llc|Hot cooling for electronics|

---

US8422218B2|2007-04-16|2013-04-16|Stephen Samuel Fried|Liquid cooled condensers for loop heat pipe like enclosure cooling|

---

CA2731994C|2008-08-11|2018-03-06|Green Revolution Cooling, Inc.|Liquid submerged, horizontal computer server rack and systems and methods of cooling such a server rack|

---

JP2012038010A|2010-08-05|2012-02-23|Fujitsu Ltd|Heat receiver, liquid cooling unit and electronic device|

---

CN203859978U|2014-05-28|2014-10-01|艾默生网络能源有限公司|Power supply module|

---

JP6710938B2|2015-11-05|2020-06-17|富士通株式会社|Data center system, control method of data center system, and program|

---

JP6768340B2|2016-04-28|2020-10-14|株式会社東芝|Railway vehicle power converter|

---

GB2569306A|2017-12-12|2019-06-19|Rolls Royce Plc|Thermal management device|

---

WO2019119137A1|2017-12-22|2019-06-27|Mcmaster University|Plate-fin heat exchanger suitable for rack-mountable cooling unit|US11152283B2|2018-11-15|2021-10-19|Hewlett Packard Enterprise Development Lp|Rack and row-scale cooling|

---

US10859461B2|2019-05-01|2020-12-08|Dell Products, L.P.|Method and apparatus for digital leak detection in liquid-cooled information handling systems|

---

CN110568911B|2019-09-05|2021-04-23|英业达科技有限公司|Server|

---

TWI695973B|2019-09-10|2020-06-11|英業達股份有限公司|Server|

---

CN110740620B|2019-10-15|2020-09-29|奇鋐科技股份有限公司|Liquid cooling distribution device used in server cabinet|

法律状态:
2017-09-20| PLFP| Fee payment|Year of fee payment: 2 |

---

2018-04-13| PLSC| Publication of the preliminary search report|Effective date: 20180413 |

---

2018-10-25| PLFP| Fee payment|Year of fee payment: 3 |

---

2019-10-24| PLFP| Fee payment|Year of fee payment: 4 |

---

2020-10-27| PLFP| Fee payment|Year of fee payment: 5 |

---

2021-10-27| PLFP| Fee payment|Year of fee payment: 6 |

优先权:

---

申请号 | 申请日 | 专利标题

---

FR1659755|2016-10-10|

---

FR1659755A|FR3057341B1|2016-10-10|2016-10-10|COMPACT LIQUID COOLING COMPACT MODULE|FR1659755A| FR3057341B1|2016-10-10|2016-10-10|COMPACT LIQUID COOLING COMPACT MODULE|

---

JP2019519222A| JP2019531556A|2016-10-10|2017-10-10|Compact liquid cooling module for computer servers|

---

RU2019113938A| RU2019113938A|2016-10-10|2017-10-10|COMPACT LIQUID COOLING MODULE OF COMPUTER SERVER|

---

US16/340,563| US11083111B2|2016-10-10|2017-10-10|Compact liquid cooling module for computer server|

---

EP17793994.9A| EP3524045A1|2016-10-10|2017-10-10|Compact liquid cooling module for an it server|

---

CA3040050A| CA3040050A1|2016-10-10|2017-10-10|Compact liquid cooling module for an it server|

---

PCT/FR2017/052782| WO2018069635A1|2016-10-10|2017-10-10|Compact liquid cooling module for an it server|

---

BR112019007125A| BR112019007125A2|2016-10-10|2017-10-10|compact computer server liquid cooling module|

---

CN201780076322.6A| CN110140437A|2016-10-10|2017-10-10|Compact liquid refrigerating module for computer server|