A cooling system can satisfy its design targets and still leave a difficult question unanswered: what did the site have to spend, consume, and change to remove the heat generated by computing? A design review may show favorable power usage effectiveness, controlled water use, and stable equipment temperatures, yet those results reveal only parts of the thermal system. The missing context sits in the interactions between electrical demand, heat rejection, water availability, ambient conditions, and the machinery required to maintain operating temperatures. When engineers optimize each part against an isolated target, the resulting system can transfer resource demand from one component to another without reducing the overall burden. Procurement teams then inherit a design that appears efficient on individual dashboards but offers an incomplete account of its total resource requirements.

The need for a unified approach does not mean that established indicators have lost their value. Power usage effectiveness, or PUE, helps operators understand how total site energy relates to information technology energy, while water usage effectiveness, or WUE, provides a way to assess water use against computing activity. Their usefulness depends on clear measurement boundaries, consistent reporting, and an understanding of what each indicator leaves outside its scope. The problem arises when procurement teams treat these complementary indicators as complete descriptions of cooling performance rather than as separate views of a larger system. A more useful framework would preserve the information they provide while examining the electricity, water, climate effects, and thermal work involved in rejecting heat. That approach would turn cooling performance from a collection of disconnected targets into an engineering assessment of the resources required to sustain computing.

Why Two Perfect Scores Can Still Mean a Poor Design

A cooling design can reach its energy target while placing greater demands on water, or protect local water resources while increasing the electricity needed to reject heat. Neither outcome automatically indicates poor engineering, because the correct balance depends on the site, the computing load, and the resources available to support operation. The problem emerges when a procurement specification rewards one result without examining the consequences for the rest of the thermal system. PUE measures the relationship between total data center energy consumption and information technology energy consumption, while WUE relates water use to information technology energy consumption. These indicators answer different questions, and neither independently establishes how efficiently the complete system transfers heat away from computing equipment and ultimately rejects it into the surrounding environment.

How Independent Targets Conceal Resource Transfers

Consider a proposed cooling design that minimizes water consumption by relying heavily on dry heat rejection. The system transfers heat to the surrounding air without depending on evaporation as its primary cooling mechanism, which can reduce direct water demand at the site. However, outdoor air temperatures determine how effectively the equipment can discharge that heat, and warmer conditions can reduce the temperature difference available to drive heat transfer. Engineers may respond by increasing airflow, selecting larger heat exchangers, raising fan speeds, or introducing mechanical refrigeration when ambient conditions exceed the system’s useful operating range. Each response changes the electrical demand of the cooling plant, even though the original decision may have improved its water performance. A procurement review that examines water consumption without testing the associated electrical requirements can therefore approve a design whose overall resource demands exceed those of a more balanced alternative.

The reverse situation can occur when a design prioritizes low electrical consumption through evaporative heat rejection. Evaporation can improve heat transfer under suitable conditions, allowing the system to reject heat without relying exclusively on energy-intensive mechanical refrigeration. That benefit comes with water requirements associated with evaporation, system maintenance, and other operating needs, depending on the chosen arrangement. The resulting energy indicator may look favorable even when the water supply faces seasonal restrictions, competing demand, or substantial treatment requirements. A lower electricity demand does not erase the consequences of withdrawing and consuming water, just as lower water consumption does not automatically justify higher electrical demand. The procurement process must evaluate both outcomes against the site’s actual resource constraints rather than treating improvement in either indicator as sufficient evidence of a superior cooling design.

Why the Heat Rejection Boundary Matters

A more complete assessment begins by following the heat rather than starting with the building’s aggregate energy balance. Electrical power consumed by processors and other information technology equipment ultimately becomes heat, which the cooling system must transfer, transport, and reject unless the site recovers some of it for useful purposes. The thermal pathway may pass through cold plates, coolant distribution equipment, secondary liquid loops, heat exchangers, chillers, dry coolers, or evaporative equipment before reaching the external environment. Each interface introduces operating requirements that depend on temperature differences, flow rates, pressure losses, control settings, and equipment performance. Measuring only the building’s energy consumption does not isolate the resources required for this thermal work, while measuring only water use cannot show whether the heat rejection process relies on additional electrical or mechanical effort.

The boundary must also distinguish between heat generated by computing equipment and heat introduced by other site loads. Electrical distribution losses, lighting, pumps, fans, and other infrastructure can contribute to the overall heat balance, but their inclusion should follow a documented accounting rule rather than an inconsistent choice made during procurement. A heat rejection assessment focused on the computing system can track the thermal load originating at information technology equipment while separately accounting for supporting infrastructure and other heat sources. This separation helps reviewers understand whether a proposed design improves the removal of computing heat or simply changes the allocation of auxiliary loads between reporting categories. It also prevents vendors from presenting different system boundaries as though their results describe the same engineering outcome. Consistent boundaries are essential before any composite measure can support meaningful comparison between competing cooling designs.

The Shift From Facility Efficiency to Heat Rejection Efficiency

The next step is to change the central question in cooling procurement from how efficiently the site operates as a whole to how effectively it manages the heat produced by computing. Facility-level indicators remain useful because they expose overheads and resource requirements that affect operating costs and environmental performance. However, those indicators do not necessarily identify which cooling architecture delivers the required thermal service with the least combined burden. A site can improve its aggregate energy result by changing information technology utilization, electrical distribution losses, or other loads without making a corresponding improvement in the cooling process itself. Similarly, a change in computing workload can alter a water-to-energy ratio even when the physical cooling plant remains unchanged. Heat rejection efficiency offers a more focused engineering perspective by relating the resources used for thermal management to the heat that the system must handle under defined operating conditions.

From Building Overhead to Thermal Work

A cooling system performs a specific physical task: it receives heat from the computing environment, transfers that heat through the required thermal pathway, and releases it to an external sink or directs it toward a useful heat recovery process. The resources needed to complete that task depend on the temperature at which the system receives the heat, the temperature at which it can reject it, and the resistance encountered along the transfer path. Higher coolant temperatures may allow some designs to reduce or avoid mechanical refrigeration, provided the computing equipment can operate safely within the resulting thermal conditions. Lower temperatures may require additional cooling effort, depending on the heat exchanger arrangement and the conditions outside the building. These engineering relationships make the thermal operating envelope a central part of any credible comparison, rather than a background specification attached to an energy target.

A heat-centered boundary can further separate the performance of the cooling architecture from unrelated changes in computing activity. When a site adds computing equipment, changes server utilization, or replaces processors with devices that produce a different heat profile, total site energy may change even if the cooling system’s underlying performance remains similar. Conversely, a cooling upgrade may reduce auxiliary energy while the computing workload stays constant, making its contribution easier to identify when the thermal load remains documented. The assessment should therefore record both the heat handled and the conditions under which the system handled it, rather than relying on an energy ratio alone. It should also identify whether the heat originates from direct liquid cooling, air-based equipment, or a combination of thermal pathways. This information gives procurement teams a stronger basis for understanding what the reported result represents and whether competing designs deliver comparable service.

Why Heat Rejection Must Include Supporting Resources

Heat rejection efficiency cannot become a meaningful procurement measure if it counts only the electricity consumed by a chiller or the fans installed beside a heat exchanger. The complete resource boundary must include the equipment that moves heat through the system, the treatment processes that condition cooling water where applicable, and the additional energy required to maintain the operating envelope. Depending on the architecture, those requirements can include pumps, coolant distribution equipment, refrigeration, heat rejection fans, water circulation, filtration, and controls. The boundary should also account for the site’s chosen redundancy strategy because standby equipment and operating configurations can change both installed capacity and actual resource consumption. A comparison that excludes these supporting loads could reward a design for moving its energy demand outside the measured subsystem rather than reducing the demand of the complete cooling process.

Water requires the same careful boundary because the amount supplied to a site does not always equal the amount consumed by its cooling process. Water may evaporate, leave through blowdown, enter a treatment system, or circulate repeatedly within a closed loop, and each pathway has different implications for local resource availability. The assessment should distinguish withdrawals from consumption and identify the source of the water, including whether the design uses potable supplies, reclaimed water, or another permitted source. Where the specification evaluates broader environmental effects, it should also state how the water requirements of electricity generation enter the accounting, because off-site electricity production can create water impacts beyond the site’s own boundary. Those categories should remain visible in the underlying report even if procurement teams later combine them into a composite result.

What Total Thermal Resource Efficiency Actually Measures

Total Thermal Resource Efficiency, or TTRE, can provide a useful structure for evaluating cooling performance if the industry defines it as a proposed composite assessment rather than an already established standard. The concept starts with a clear question: what resources does a site require to manage a defined thermal load while maintaining the operating conditions that computing equipment needs? Answering that question requires a consistent account of electricity, water, climate-related impacts, and the heat handled by the cooling system. The assessment must also recognize that these resources interact, because a reduction in one category can increase demand in another. A composite measure would bring those relationships into a common evaluation while preserving the underlying results for engineering review. Its purpose would be to support better design and procurement decisions, not to replace physical measurements with an opaque score.

Defining the Composite Without Hiding Its Components

A useful TTRE model needs a functional definition before procurement teams can compare bids or establish acceptance criteria. The proposed boundary should begin at the computing equipment’s thermal interface and follow the heat through the distribution system to the final heat sink or a verified heat recovery destination. Within that boundary, the assessment would record the electricity required by cooling equipment, the water withdrawn and consumed by the relevant processes, and the thermal load managed under the specified operating conditions. It would also document the site’s environmental context, including the carbon intensity of the electricity supply and the local consequences of water demand. Those inputs describe different physical quantities, so the methodology cannot combine them responsibly through an unexplained addition or an arbitrary weighting scheme.

Cooling Water Efficiency Opportunities for Data Center

The thermal load provides the functional reference that allows the other inputs to be interpreted together. Engineers should document the heat transferred from the computing equipment, the temperature conditions at the relevant interfaces, and the quantity of heat rejected or usefully recovered during the assessment period. Electrical consumption should include the pumps, fans, refrigeration equipment, coolant distribution systems, water treatment, and controls that support that thermal service. Water accounting should separately identify withdrawal, consumption, source, treatment requirements, and relevant discharge pathways rather than treating all water movement as equivalent. Climate impact should follow an explicitly defined method that reflects electricity-related emissions and any additional impacts the specification intends to include. Keeping these inputs visible allows reviewers to identify why one design performs differently from another, even when both report similar overall results.

A composite assessment must also avoid giving the appearance that unlike resources can always compensate for one another. A lower electricity requirement cannot automatically justify excessive pressure on a scarce local water source, while reduced water consumption cannot automatically excuse a large increase in electricity demand. Procurement teams need to define acceptable resource limits and environmental constraints before applying any weighting or normalization method. A credible TTRE model could use a composite score for ranking designs within an approved operating envelope, while retaining separate thresholds for water stress, energy demand, emissions, and thermal performance. This arrangement would prevent a strong result in one category from concealing an unacceptable result in another. It would also give engineering teams enough detail to explain the trade-offs behind the final ranking rather than asking decision-makers to trust a single unexplained figure.

Establishing a Defensible Calculation Method

The calculation method should distinguish direct measurements from modeled inputs and disclose the assumptions used when equipment data cannot provide a complete picture. Cooling electricity can often be measured through dedicated meters, while water flows require appropriately located meters and a clear distinction between makeup supply, recirculation, and discharge. Thermal measurements need to capture the relevant temperatures and flow conditions so that the assessment can determine how much heat the system transfers under the specified operating envelope. Where the calculation estimates indirect emissions or off-site resource impacts, the methodology should identify the electricity mix, water source assumptions, and applicable accounting boundaries. Procurement teams should require vendors to provide the data behind every material input rather than submitting a final score without supporting evidence.

Normalization should follow the physical purpose of the comparison rather than an arbitrary desire to produce one universal number. A design handling a different thermal load, maintaining a different coolant temperature, or operating under different ambient conditions cannot be compared fairly with another design unless the method accounts for those differences. Engineers should therefore define the reference thermal service, workload conditions, measurement period, and operating limits before calculating the composite result. Where seasonal performance matters, the calculation should use a representative range of conditions rather than relying exclusively on a convenient test window. Where redundancy changes the operating configuration, the methodology should state whether the assessment represents normal operation, a specified equipment outage, or another defined condition. These rules give TTRE a consistent technical basis while allowing the result to reflect the operating circumstances that actually shape cooling demand.

Why the Same Efficiency Number Means Different Things in Different Climates

Cooling performance depends on the conditions surrounding the site as much as on the equipment installed inside it. Outdoor temperature affects the ability of dry coolers and heat exchangers to reject heat, while humidity influences evaporative cooling potential and the conditions under which moisture can leave the system. Water availability changes the significance of designs that depend on evaporation, even when those designs offer favorable electrical performance. The local electricity supply also shapes the environmental implications of cooling demand, because identical electricity consumption can produce different emissions depending on how power is generated. A site-independent score would overlook these differences and could rank designs in a way that fails to reflect their actual resource consequences. TTRE therefore needs a climate-aware method that evaluates equipment performance under defined local conditions without allowing the location itself to excuse poor engineering.

Ambient Conditions Change the Engineering Outcome

Ambient temperature shapes the temperature difference available for transferring heat from a cooling loop to the surrounding environment. When outdoor conditions remain favorable, a dry cooler may reject heat with limited mechanical assistance, provided the system can maintain the required coolant temperature. As outdoor temperatures rise, the same equipment may need higher fan speeds, greater heat exchanger capacity, or supplementary refrigeration to preserve the operating envelope. These changes can increase electrical demand even though the installed equipment and nominal design remain unchanged. A climate-aware comparison should therefore evaluate the system across the weather conditions relevant to the site’s expected operation, including periods when heat rejection becomes most demanding. This assessment helps procurement teams distinguish a design that performs well under mild conditions from one that maintains reliable thermal service when the local climate places greater pressure on the cooling plant.

Humidity introduces another set of conditions that affects the relative performance of cooling architectures. Evaporative systems depend on the air’s capacity to accept additional moisture, so their performance changes with the relationship between dry-bulb temperature and wet-bulb temperature. In suitable conditions, evaporation can support heat rejection at temperatures that would otherwise require more mechanical cooling. In humid conditions, the available evaporative advantage can narrow, changing the balance between water consumption and electrical demand. Engineers must also consider the operating limits of the selected equipment, water quality requirements, and any controls that change the cooling mode as weather conditions shift. TTRE should capture these changes through documented climate inputs and operating profiles instead of treating humidity as a secondary design detail that has no bearing on the procurement result. 

Water Availability and Electricity Supply Must Shape the Result

Water availability is a local constraint that cannot be represented adequately by the volume used alone. A given amount of consumption may have different consequences depending on the source, competing demand, seasonal availability, treatment requirements, and the condition of the relevant water basin. A site supplied by reclaimed water may have a different local impact from one that relies on a constrained potable water source, although reclaimed supplies still require assessment of availability, quality, treatment, and competing uses. Procurement teams should require the proposed TTRE method to identify these factors and explain how they affect the evaluation. The model should not assign an automatic environmental advantage to a water source simply because it carries a particular label. A defensible result must connect the water accounting to the local conditions that determine whether the cooling system’s demand is sustainable and operationally secure.

Electricity supply creates a parallel challenge because the same cooling load can have different climate implications across locations and operating periods. A site drawing electricity from a lower-carbon supply may report a different emissions impact from a site with a more carbon-intensive supply, even when both cooling systems consume comparable amounts of electricity. Grid conditions can also affect the operational value of reducing cooling demand during periods of constrained supply. A climate-aware TTRE assessment should identify the emissions factors used, their geographic and temporal relevance, and whether the method evaluates average or marginal electricity impacts. It should avoid mixing these approaches without disclosure, because each answers a different question about the consequences of electricity use. By separating physical energy consumption from its environmental interpretation, the assessment can support consistent engineering comparisons while making the location-specific consequences visible to procurement decision-makers.

How to Write This New Efficiency Into an RFP Without Creating Loopholes

A cooling performance requirement becomes useful only when a procurement document defines exactly what vendors must measure, report, and demonstrate. Introducing TTRE without a shared boundary or calculation method could create another layer of inconsistent reporting instead of resolving the limitations of isolated PUE and WUE targets. Procurement teams should begin by describing the thermal service required, the equipment included in the cooling system, the resources counted, and the environmental conditions used for comparison. They should then specify the proposed composite methodology, the underlying measurements, the permitted assumptions, and the evidence required to verify each result. Separate resource constraints must remain visible so that a favorable composite score cannot conceal excessive water demand, unacceptable electrical requirements, or inadequate thermal performance.

Define the Boundary Before Setting the Target

An effective RFP should identify the point where the assessment begins and the point where it ends, including every material component that supports heat transfer and rejection. The boundary should address the computing equipment interface, coolant distribution, pumps, heat exchangers, refrigeration equipment, heat rejection systems, controls, and water treatment where these components form part of the proposed cooling solution. The document should explain how the calculation treats shared equipment, standby systems, redundancy, and supporting loads that serve more than one part of the site. Where electrical or water infrastructure sits outside the vendor’s immediate equipment package but materially supports the cooling design, the RFP should require disclosure of its associated resource demand. These rules prevent a bidder from improving a reported result simply by moving an energy-consuming component beyond the stated measurement boundary.

The RFP should then specify the TTRE methodology as a documented calculation rather than a target number detached from its inputs. Procurement teams should require the formula, units, weighting assumptions, normalization rules, data sources, and treatment of missing measurements before evaluating bids. Vendors should provide separate electricity, water withdrawal, water consumption, climate impact, and thermal load results alongside the proposed composite score. The document should define the conditions under which the parties may use modeled data and identify when measured operating data must replace estimates. It should also establish how reviewers will verify the results through equipment schedules, meter locations, calculation files, and commissioning records. By setting these requirements before bid submission, procurement teams can reduce the scope for selective reporting and ensure that every bidder evaluates the same thermal service against the same accounting rules.

Build Verification Into the Contract

A robust procurement clause should distinguish the design-stage estimate from the performance demonstrated after installation. Vendors can model expected performance using equipment specifications and assumed operating conditions, but those projections do not establish how the complete system will perform once installed and commissioned. The contract should therefore define the measurements required during acceptance testing, including electrical consumption, water flows, thermal conditions, and the operating state of the relevant equipment. It should specify the duration and representativeness of the test, the instrumentation requirements, and the method for resolving missing or inconsistent readings. Where seasonal conditions prevent a complete test at handover, the parties should agree on a documented procedure for validating performance under additional operating conditions. This structure makes the procurement commitment measurable without treating a short commissioning test as proof of performance across every climate or workload condition.

The measurement plan should identify which instruments provide the data used in the composite calculation and how the parties will maintain their accuracy. Electrical meters must capture the relevant cooling loads rather than relying on estimates derived from equipment nameplates or assumed utilization. Water meters should distinguish makeup supply and relevant discharge streams where necessary to determine withdrawal and consumption accurately. Temperature and flow measurements should support the thermal calculation, while the environmental inputs should come from documented sources that match the agreed reporting period and location. The contract should specify how the parties will handle sensor failures, data gaps, equipment changes, and deviations from the expected operating envelope. These provisions reduce the risk that different measurement practices will produce apparently comparable results that actually describe different physical conditions.

When Efficiency Stops Being a Dashboard and Starts Driving Design

A composite resource assessment can influence cooling architecture before engineers commit to major equipment selections, pipework, electrical capacity, and site infrastructure. When procurement teams evaluate electricity, water, climate impact, and thermal performance together, the design process must account for the consequences of decisions across the entire heat rejection pathway. Engineers can then assess whether a particular coolant temperature reduces refrigeration demand, whether a different heat exchanger arrangement increases fan energy, or whether a water-saving design creates additional mechanical requirements. The value lies in exposing these relationships while the design remains flexible enough to change. A dashboard records the consequences of decisions after engineers make them, whereas a well-defined TTRE requirement gives the design team a common basis for evaluating those decisions before they become expensive to reverse.

Thermal Architecture Becomes a Resource Decision

The choice between air cooling, direct liquid cooling, and hybrid arrangements affects the temperatures, flow requirements, and heat transfer interfaces that shape the rest of the system. Direct liquid cooling can collect heat close to the components that generate it, but the resulting architecture still needs to move that heat through coolant distribution equipment and reject it through an appropriate external system. Air cooling may simplify some liquid distribution requirements while demanding substantial airflow and heat exchange capacity under demanding conditions. Hybrid systems can combine these approaches, although their performance depends on how effectively the different loops operate together and how the controls manage changing loads. A TTRE assessment should compare these architectures against equivalent thermal requirements rather than assuming that a particular technology guarantees a better overall result.

Coolant temperature deserves particular attention because it influences the amount of mechanical work required to move heat from the computing equipment to the external environment. Raising the permitted coolant temperature may expand the opportunities for economization or direct heat rejection, provided the computing equipment remains within its approved operating limits. Lower coolant temperatures can increase the demand for refrigeration when outdoor conditions do not provide a sufficient temperature difference for heat rejection. Engineers must evaluate those effects alongside flow requirements, pressure losses, heat exchanger performance, and the operating limits of the complete thermal chain. A composite resource requirement encourages the design team to examine whether each temperature choice reduces total resource demand or simply shifts consumption between refrigeration, pumping, and heat rejection.

Materials, Heat Reuse, and Long-Term Adaptability

Material selection can influence thermal performance through heat transfer properties, corrosion resistance, compatibility with the chosen coolant, and the maintenance requirements of the system. A material that performs well under initial operating conditions may require additional treatment or more frequent intervention if the coolant chemistry, water quality, or operating environment creates compatibility problems. Procurement specifications should therefore ask vendors to disclose the material and fluid assumptions that underpin their performance calculations, particularly where those choices affect heat transfer or operating reliability. The evaluation should avoid assigning a universal environmental advantage to a particular material without considering its application, service life, and maintenance requirements. Integrating these factors into the resource assessment can reveal trade-offs that a component-level efficiency specification would otherwise overlook.

Heat reuse creates another design decision because the temperature of rejected heat determines whether a nearby process can use it directly or requires additional equipment to raise its temperature. A cooling system that supplies heat at a suitable temperature may support a practical recovery arrangement, while a lower-temperature stream may require a heat pump and additional electricity before it becomes useful. The design must also account for the distance between the source and recipient, the availability of a consistent heat demand, and the timing of that demand relative to computing operations. A connection that has no reliable recipient should not receive the same credit as a recovery system that demonstrably displaces another source of useful heat. TTRE can incorporate this distinction by documenting recovered heat, the resources required to deliver it, and the evidence that it serves a real demand.

From Cooling Performance to Thermal Accountability

The move beyond isolated PUE and WUE targets begins with a more complete understanding of what cooling systems actually do and what they require to do it reliably. Electricity, water, climate impact, and thermal load interact throughout the heat rejection pathway, so a change that improves one indicator can create additional demands elsewhere. A procurement process that evaluates these resources independently may overlook those transfers and reward designs that perform well against individual targets without delivering the best overall resource outcome. Total Thermal Resource Efficiency offers a proposed structure for examining those relationships through a consistent thermal boundary, documented measurements, and an explicit calculation method. Its value depends on preserving the underlying resource data, establishing clear constraints, and validating the results against the operating conditions that matter to the site.

The industry would need to establish common definitions before TTRE could support consistent comparisons across sites, cooling architectures, and procurement programs. Those definitions should cover the thermal boundary, measurement requirements, treatment of indirect impacts, climate normalization, weighting assumptions, and independent verification of reported results. Separate resource limits must remain part of the evaluation so that a favorable composite score cannot conceal unacceptable water demand, electrical requirements, or operating conditions. Procurement teams should also distinguish measured performance from modeled projections and document how changes in workload, weather, or equipment configuration affect the reported result. A shared methodology would make comparisons more transparent, but it would not remove the need for engineering judgment or eliminate the influence of local constraints.