Designing and implementing a large-scale automated –80°C archive
The Automation Partnership, York Way, Royston, Hertfordshire, SG8 5WY, UK.
* Corresponding author. E-mail: justin.owen{at}automationpartnership.com
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Background This article describes an automated facility for storing biological samples at –80°C, designed to meet the needs of The UK Biobank. No store of this size has previously used liquid nitrogen as refrigerant, and so it was important to confirm the storage compartments could achieve and maintain uniform conditions with efficient use of the coolant. In addition, the store incorporates a novel system of drawers designed to allow robotic access whilst maintaining storage conditions. Experiments were undertaken to prove the performance of both these factors in maintaining the temperature of stored samples, both in steady-state conditions and during sample access.
Methods Sample tubes containing water were placed in key locations within a prototype storage compartment. Thermocouples inside these sample tubes were used to together with automated data loggers to accumulate continuous records of the temperature of the samples during the experiments.
Results Results show that the design of refrigeration system and storage compartment, using LN2 as refrigerant, enables efficient use of coolant and maintains uniform temperature over the entire compartment. The results also show that samples within the compartment remain at the intended storage temperature during automated retrieval operations, including the situation when an access drawer remains open indefinitely.
Conclusions These results confirm that the store design meets the requirements for the UK Biobank, and have given both the manufacturer (The Automation Partnership) and the UK Biobank the confidence to proceed with the implementation of a full sized automated store to hold 10 million samples.
Keywords Automated biobanking, automation for biobanking, biological sample management, Polar, –80°C sample storage, ultra low temperature (ULT) storage
Accepted 10 December 2007
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
The UK Biobank initiative1 requires a repository of an unprecedented scale to maintain an inventory of samples, from up to 500 000 participants, stored in support of research for the long-term health benefit of the UK population. As part of this, an automated archive will provide storage for 10 million samples held at –80°C in microtubes with an intended useful life of up to 25 years.
The creation and operation of such a repository presents a range of challenges; the design of environmental control systems to maintain storage conditions continuously over the lifetime of the project, robotic access to samples and supporting a substantial level of storage and retrieval transactions whilst ensuring that the inventory model remains current and error free.
These challenges mirror those surmounted by The Automation Partnership (TAP) in the scale up of compound management since the 1990s2 and more recently in the automated processing of large DNA collections.3 A typical large pharmaceutical compound library will comprise several million dissolved samples which are randomly accessed to identify and confirm early stage drug candidates in high-throughput screening. The scale of such collections and the nature of accessing the samples are both comparable to those involved in the UK Biobank project. TAP has considerable expertise in supplying innovative and robust large automated storage and retrieval systems to customers in the life science sector and was selected by the UK Biobank to design and implement the –80°C automated sample management system, POLARTM.
Key factors for the successful operation of the automated store are as follows:
- Continuity of storage conditions
- Robustness of operation
- Integrity of sample tracking and accuracy of identity
- Sample security
- Cost of ownership
This article describes the main features of POLAR (Figure 1). Many aspects of the design are unique, offering previously unattainable levels of performance and security.
|
Choice of refrigeration technology
Mechanical refrigeration is the only option in widespread use, either in the form of ‘freezer farms’—discrete manually accessed units, or as large centralised complex refrigeration systems serving insulated chambers within sealed enclosures. Both these approaches are reliant on cascade refrigeration, which has many disadvantages when applied to large scale sample storage.
Limited temperature and cooling capacity
Conventional single compressor refrigeration systems are only capable of achieving temperatures of about –40°C. If lower temperatures are required then cascade refrigeration systems must be used. A two stage cascade system uses two refrigeration circuits, a ‘high stage’ and a ‘low stage’, connected in series. Two different refrigerants are required. The low stage is used to cool the process using a very low boiling point refrigerant. The high stage cools the low stage, removing both the process heat and the work done by the low stage compressor and other heat inputs. For large amounts of heat transfer such systems are complex and inefficient and often require water cooling.
Temperature of –80°C is very close to the limit of two-stage cascade refrigeration systems and the cooling capacity at such low temperatures is severely limited. Consequently, it takes significant time to establish and to restore the required temperature after opening a compartment for access, especially after a fault such as a door being left open. Furthermore, as there is little margin for loss of performance, the cooling systems and insulated chambers must be kept in optimum condition throughout their service life.
Maintenance costs
Cascade systems require regular costly maintenance due to the large number of moving parts and high temperatures and pressures involved. Servicing and repair requires specialist and often proprietary knowledge that may not be available over a 25-year period. Typically a liquid nitrogen system has between one and two orders of magnitude fewer working mechanical/electrical parts than a large cascade refrigeration-based solution. This greatly simplifies spares holding, reduces the risk of obsolescence and generally makes the system much easier to support over the long term.
Energy costs
Cascade refrigeration is extremely inefficient and costly to run. Expensive backup generators are also required to provide security in the event of mains power loss.
Building services requirements
Cascade refrigeration generates large quantities of waste heat, placing a demand on air conditioning plant which must be upgraded accordingly. For larger installations expensive water cooling systems are often required. The compressors are noisy and sound attenuation is necessary to keep sound levels within safe working limits.
After evaluating mechanical refrigeration, TAP and the UK Biobank explored the viability of using liquid nitrogen (LN2) as a refrigerant and this was chosen as the preferred approach. The scheme, in which LN2 is injected into heat exchangers located in the sample storage chambers, has many advantages, principally:
- No compromise on storage temperature—liquid nitrogen boils at –196°C.
- Fast cool down and recovery rates—cooling capacity is effectively unlimited compared to mechanical systems.
- Reliability—intrinsically simple design with no highly stressed components and very few moving parts, LN2 is self pumping.
- Ease of maintenance—no moving parts or serviceable items in the ULT compartments.
- Efficiency—temperature of the waste gas is only a few degrees above the sample storage temperature.
- ‘Building friendly’—no waste heat and quiet in operation.
- For UK Biobank, using liquid nitrogen to cool the archive instead of electricity had the additional advantage of a significant cost saving. However, this is dependent upon local utilities costs and should be analysed on a case by case basis.
Although the advantages of liquid nitrogen were compelling, the performance of such a system was unproven, consequently it was necessary for TAP to develop and test this design.
In a conventional refrigerator samples are accessed by opening a large door, resulting in the inflow of warm air and moisture. In this case, due to the risk of damage to samples, access times are limited and the time between placements or withdrawals must be carefully managed to ensure that the freezer contents can recover to –80°C. These constraints limit sample acquisition rates. Failure to close a door due either to operator error in a manual facility or as the result of an exception in an automated store may result in the damage or loss of a large number of samples.
We have developed a novel design for sample access that overcomes these limitations and is supported by a patent application.4 ULT compartments are separated from a highly de-humidified –20°C zone by fixed insulated panels except for the face where robotic access is required. Storage drawers fitted with insulation ‘tile’ the access faces, forming an insulating wall and creating a sufficiently good seal to allow the refrigeration system to comfortably maintain conditions. Each compartment contains over 500 drawers, each of which can be individually accessed by the aisle robot. Less than 0.2% of the area is open to the aisle at any time while archiving or accessing samples, limiting the number of samples exposed and preventing collateral warming of adjacent drawers.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
The principal areas of development were the refrigeration of the –80°C storage compartments and robotic access to the samples held within them. TAP, with UK Biobank's support, developed and commissioned a fully functional ultra low temperature (ULT) compartment, including a full-size storage frame and sample storage drawers, in the early stages of the project. Having the facility at TAP made it possible to gather real data on the store performance and prove many aspects of the design.
To determine temperature stability and uniformity, two Grant SQ800 dataloggers, configured to sample the temperature of 14 thermocouples located in a ULT compartment were used to record 42 000 values over a 50 h period.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
Figure 2 shows the data from initial cool down to 40 h of steady-state operation. The mean temperature of the ULT compartment, calculated by averaging temperature of each of the 12 thermocouples over the 40 h period was –82.2°C with a standard deviation of <1.6. The minimum and maximum temperatures recorded over this period were –87.8°C and –79°C, respectively.
|
The purpose of measuring the evaporator inlet temperature was to confirm that liquid nitrogen was present. If the LN2 boiled off before reaching the evaporator then the cooling efficiency would be much reduced. There is only a small temperature difference between the exhaust gas and the mean temperature in the compartment (Figure 2), demonstrating that the evaporator is extremely efficient.
Figure 3 demonstrates that only samples on the opened drawer experienced a change in temperature. Furthermore, during a series of tests to simulate various failure conditions, the samples on adjacent drawers remained at –80°C when the centre drawer was left open indefinitely.
|
Therefore, no recovery time is required for the compartment as its temperature is unaffected by accessing the drawers. This is a key factor in achieving high sample picking rates and maintaining sample integrity.
In this test, the temperature of an isolated tube increases to around –65°C. This is worst case as fully populated racks are warmed more slowly due to their greater thermal mass and reduced convective transfer due to the close packing of the tubes within the rack.
The recovery time for a single tube in a rack is 
20 min. However, the scale of inventory relative to daily picking rates and the random distribution of samples within the store ensures that the probability of returning to a drawer to pick another sample within this time is extremely low, and the store control software can, if necessary, easily avoid such coincidences. Therefore, the throughput of the store is not impeded by sample recovery time.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
The main purpose of this article has been to demonstrate how POLAR has been designed to ensure continuity of storage conditions and sample integrity. As a result the emphasis has naturally been on the development of the storage compartments and refrigeration but there are other important factors:
Robustness of operation
POLAR includes features developed and refined over the course of many successful projects to ensure that the impact of operator errors, robotic faults, power outages and component failures in computers, control systems and refrigeration plant are unlikely to threaten the security of stored samples.
Integrity of sample tracking and accuracy of identity
Store robotics check the barcode of each microtube every time it is moved ensuring 100% verification of identity and allowing vessel movements to be tracked and logged without errors. No manual system of this scale has ever been shown to achieve this.
Security of access
POLAR does not require operators to approach the stored inventory, which lies within a locked enclosure to which access is restricted. Operator access to the store user interface is password protected. Orders for sample retrieval can only be generated through the Laboratory Information Management System (LIMS) and cannot be instigated by a store operative.
Cost of ownership
POLAR should prove more cost-effective in the long term as only one operative is required to support the large number of store transactions required. To achieve similar rates manually a much larger group of trained laboratory staff would be necessary. The design of Polar means that samples can be retrieved as fast as the robotics is able to operate. The store can collect and collate sample tubes dispersed throughout a stored inventory of 10 million at a rate of about 100 per hour. The expected actual loading on the store will be well within this capability. The economics of the automated solution become even more compelling if the cost of errors and lost or damaged samples, which would be inevitable in a manual repository of this scale, are considered.
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
In this article we have described a novel architecture for automated storage of samples at –80°C, which has been developed to meet the needs of The UK Biobank. In a system of this scale, liquid nitrogen-based refrigeration has significant advantages over mechanical refrigeration. The performance of a liquid nitrogen cooled store has been evaluated through tests run on a full-size pilot compartment, demonstrating that excellent temperature stability and uniformity can be achieved. Further tests have shown that the same design allows individual sample drawers to be accessed robotically without impacting the temperature distribution within the compartment.
Conflict of interest: J Owen and P Woods are employees of, and also shareholders in, The Automation Partnership.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionsReferences |
|---|
1 Elliott P, Peakman TC. The UK Biobank sample handling and storage protocol for the collection, processing and archiving of human blood and urine. Int J Epidemiol (2008) 37:234–44.
2 Archer JR. History, evolution and trends in compound management for high throughput screening. ASSAY Drug Dev Technol (2004) 6:2.
3 The Automation Partnership. Managing Libraries of DNA Samples-How TAP's automation enables scale-up in Genomics research (2006) . White paper available from The Automation Partnership on request.
4 European Patent Application No. 05108316.0 The Automation Partnership (Cambridge) Limited.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Elliott, T. C Peakman, and on behalf of UK Biobank The UK Biobank sample handling and storage protocol for the collection, processing and archiving of human blood and urine Int. J. Epidemiol., April 1, 2008; 37(2): 234 - 244. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. A Manolio Biorepositories--at the bleeding edge Int. J. Epidemiol., April 1, 2008; 37(2): 231 - 233. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||