Which Cooling Medium is Best for the Direct-to-Consumer Food Delivery Market?

Key Takeaways:

When choosing a cooling medium for the direct-to-consumer food delivery, it must keep its payload to specification upon arrival in any temperature condition over time and by any means of transportation with a minimum of a 95% success rate to be considered a viable option. Through our testing of dry ice, phase change materials (PCMs), and gel packs, dry ice not only proved to be the least costly, but also the highest performing cooling medium with the lowest mass in any container type.

Cooling Mediums in the Direct-to-Consumer Food Delivery Market

The meal kit and direct-to-consumer food delivery market has grown into a significant logistics challenge. Companies in this space ship perishable, temperature-sensitive products such as proteins, dairy, produce, and prepared meals directly to consumers, often across complex multi-leg routes that combine warehousing, road transport, and last-mile delivery.

Unlike traditional retail cold chain where product moves through temperature-controlled environments for most of its journey, direct-to-consumer food shipments spend extended periods in passive packaging with no active refrigeration. The payload arrives in specification or it does not. There is no intervention point in between.

The commercial consequences of a temperature excursion in this model are immediate and compounding. A single failed shipment generates a customer service cost, a product replacement or refund, and a reputational event that is disproportionately damaging in a subscription-based business where retention is the primary economic driver. At scale, even a low failure rate represents a meaningful operational liability. A 5% monthly failure rate across 500 shipments is 25 lost orders every month, on a single lane.

Solving this problem reliably requires getting two interdependent decisions right: the insulated container and the cooling medium inside it.

Rethinking the Cooling Medium Decision

A temperature-sensitive shipment depends on the container and the cooling medium to function as an integrated system. Container wall thickness, insulation material, and payload geometry determine how quickly heat transfers into the package. The cooling medium determines how long the system can absorb that heat load before the payload temperature drifts outside specification. Get either decision wrong and the system fails, regardless of how well the other component performs.

In practice, both decisions are often made with less rigor than the stakes warrant. Container selection tends to follow vendor relationships or established procurement patterns. Cooling medium selection between dry ice, phase change materials (PCMs), or water-based gel packs is frequently based on habit, prior qualification results, or cost-per-unit comparisons that do not account for how performance varies across lanes, seasons, or ambient conditions. A configuration that passes laboratory qualification may still fail on a demanding summer route. One that performs reliably on a short UK ground lane may be inadequate for a cross-continental US air lane.

This blog presents a data-driven framework for cooling medium selection grounded in thermodynamic simulation and historical lane temperature data across representative US and UK shipping routes. The framework treats packaging and cooling mediums as the integrated system they are.

Additionally, the research focuses specifically on the cooling medium decision where the performance differences between options are largest and where the financial and environmental consequences of a poor choice are most significant. The analysis compares dry ice, PCM, and gel pack solutions across three dimensions that matter to technical and logistics decision-makers: thermal reliability, total cost of ownership, and carbon footprint. All dimensions utilize payload temperature requirements of 32°F (0°C) to 41°F (+5°C).

The Three Tested Cooling Mediums

Before comparing performance, it is worth establishing what each cooling medium is doing thermodynamically because the mechanisms differ significantly, and those differences drive real-world outcomes.

Dry Ice (Solid CO₂)

Dry ice sublimates at -109.3°F (-78.5°C), transitioning directly from solid to gas without passing through a liquid phase. This extremely low phase-change temperature gives it high cooling capacity per unit mass, which is why it can maintain a payload within a narrow range like 32°F (0°C) to 41°F (+5°C) using relatively small quantities. Typical dry ice amounts range from 3.3 lb (1.5 kg) to 4.4 lb (2.0 kg) for a 33L container over a 24-hour lane.

The challenge is that dry ice does not modulate—it sublimes continuously. If product and dry ice are in direct thermal contact without a buffer layer, cold excursions below 32°F (0°C) are a risk that must be managed through container design. A corrugated cardboard pad placed between the product and the dry ice bricks is the standard mitigation. Container wall thickness also plays a meaningful role; thicker walls slow heat ingress, reducing the sublimation rate and extending the effective cooling window.

Phase Change Materials (PCMs)

PCMs, typically organic compounds formulated to transition at a specific target temperature, absorb and release energy at their phase change point rather than across a continuous temperature gradient. A +35.6°F (+2°C) PCM maintains its phase change temperature as long as the material is transitioning to provide a stable thermal buffer directly aligned with the payload’s allowable range.

The tradeoff is mass: effective PCM configurations for a 24-hour lane typically require 6.6 lb (3.0 kg) to 11.0 lb (5.0 kg) of coolant, which is two to three times the mass of a dry ice configuration with comparable reliability. Container insulation performance matters significantly with PCMs. A better-insulated container reduces the rate at which the PCM must absorb heat, extending its effective duration, and in some cases, allowing a reduction in required coolant mass.

Water-Based Gel Packs

Gel packs use a water-based medium with a phase change at approximately 32°F (0°C). They are inexpensive, widely available, and familiar to the industry. However, their phase change temperature sits at the lower bound of the allowable payload range, which creates a narrow margin for thermal management. Adequate gel pack configurations for demanding lanes, particularly routes with high ambient temperatures or significant seasonal variation, often require substantial coolant mass of 6.6 lb (3.0 kg) or more.

Container insulation quality is especially critical with gel packs. A lower-performing container compresses the available thermal buffer and accelerates failure during high-ambient conditions. The thermal dynamics of gel packs tend to perform well in cooler ambient conditions and deteriorate significantly in warm-weather months, as the lane analysis that follows demonstrates. 

How Cold Chain Lane Risk is Measured

Laboratory qualification tests establish whether a packaging system can maintain payload temperature within specification under a defined set of conditions. What they cannot establish is how that system will perform across the full distribution of ambient temperature conditions that a real shipment will encounter over the course of a year.

The analysis presented in this paper uses a daily lane-risk approach developed by SmartCAE's Digital Cold Chain (DCC) simulation platform. The methodology works as follows:

An accurate thermodynamic model of each shipping container is constructed using material properties, wall thickness, payload specifications, coolant type and mass, and initial temperatures for all components. Lane-specific ambient temperature profiles are generated by pulling historical weather station data along each route segment, which includes solar irradiance for segments where the container is exposed to direct sunlight, such as tarmac time, loading docks, and last-mile delivery. A separate temperature profile is generated for every day within a three-year historical window (2021–2023), producing over 850 profiles for the US lane and over 1,026 profiles for the UK lane.

The thermodynamic simulation is then run for each profile, testing whether the payload temperature remains within the allowable range for the entire lane duration. The output is a daily pass/fail result that aggregates into a monthly and annual failure rate to provide a statistically robust predictor of real-world performance across the full range of conditions a shipper will encounter on that lane.

The threshold for an acceptable solution in this analysis is a monthly pass rate of at least 95%. Any configuration that fails more than 5% of profiles in any given month does not qualify as a reliable solution for year-round deployment on that lane.

Cooling Medium Performance Across Real Shipping Lanes

The analysis covers two lanes selected to represent contrasting cold chain scenarios. The first is a 24-hour ground route from London to Glasgow, representative of short-haul domestic delivery in a temperate climate. The second is a 24-hour air/ground route from Richmond, VA to San Francisco, CA, representative of a more thermodynamically demanding cross-country lane that combines road segments with a commercial flight and exposes the payload to a wider range of ambient conditions. Together, the two lanes illustrate how cooling medium requirements shift as route complexity and seasonal temperature variation increase.

Each configuration tested pairs a container insulation material with a specific cooling medium and mass. Configuration codes follow a standardized convention: EPS denotes expanded polystyrene, SUS denotes sustainable composite material, DI denotes dry ice, IP denotes gel pack, and PC denotes phase change material. The numeric suffix indicates coolant mass. For example, EPS-DI01.5kg denotes an EPS container with 3.3 lb (1.5 kg) of dry ice. Only configurations meeting the 95% monthly pass rate threshold are carried through to the cost and carbon analysis that follows.

Lane 1: London, UK to Glasgow, UK (24-Hour Ground, Refrigerated 32–41°F / 0–5°C)

The UK lane covers a 24-hour ground route structured across five segments: packaging and loading, warehouse pickup, road transit, destination warehouse, and last-mile delivery. Analysis draws on 1,026 daily temperature profiles across three years.

Dry Ice Results:

The most important finding on this lane is the sensitivity of dry ice performance to coolant mass. At 2.2 lb (1.0 kg) of dry ice in an expanded polystyrene (EPS) container, the overall pass rate across all 1,026 profiles is 84.5%, driven into failure territory during summer months. The monthly failure rate reaches 36.5% in June, 64.4% in July, and 52.2% in August. A shipper running 500 shipments per month on this lane during July using this configuration should expect roughly 322 failed shipments, a commercially unacceptable outcome.

Increasing dry ice mass to 3.3 lb (1.5 kg) eliminates this risk entirely. Both EPS and sustainable material (SUS) containers with 3.3 lb (1.5 kg) of dry ice achieve a 100% pass rate across all 1,026 profiles and all 12 months. This 1.1 lb (0.5 kg) increment in coolant mass is the difference between a solution that works reliably year-round and one that fails at scale during peak ambient temperature months.

Phase Change Materials Results:

Both EPS and SUS containers with 6.6 lb (3.0 kg) of +35.6°F (+2°C) PCM achieve 100% pass rates across all profiles and all months on this lane. PCM is a reliable solution at this configuration, but it requires twice the coolant mass of the compliant dry ice configuration and a heavier container to accommodate it.

Gel Packs Results:

Gel pack configurations achieve 100% pass rates at 4.4 lb (2.0 kg) in EPS and 6.6 lb (3.0 kg) in SUS containers on this lane. Gel packs are viable on the UK lane, but require more mass than dry ice to achieve equivalent reliability. The SUS container's slightly lower insulation performance — thermal conductivity of 40 mW/m·K versus 35 mW/m·K for EPS — requires an additional 2.2 lb (1.0 kg) of gel pack coolant to compensate, illustrating the direct interaction between container and cooling medium performance.

Lane 2: Richmond, VA to San Francisco, CA (24-Hour Air/Ground, Refrigerated 32–41°F / 0–5°C)

The US lane is more thermodynamically demanding. It combines road segments in Virginia and California with a six-hour flight from Washington Dulles to San Francisco, creating a more complex ambient temperature profile that includes tarmac exposure and high-ambient road segments in warmer months. Analysis draws on 859 daily profiles.

Dry Ice Results:

The US lane clearly reveals the failure modes of under-specified dry ice configurations. The EPS-DI01.0kg configuration fails severely during warmer months: 61.6% failure in May, 100% failure from June through September, and residual failures in October and November. A 1.1 lb (0.5 kg) increase to 3.3 lb (1.5 kg) of dry ice resolves the problem entirely — both EPS and SUS containers with 3.3 lb (1.5 kg) of dry ice achieve 100% pass rates across all 859 profiles year-round. As on the UK lane, the interaction between container insulation performance and dry ice sublimation rate is a meaningful variable: the thicker-walled UK containers (40 mm) versus the thinner US containers (25 mm) produce different sublimation dynamics that must be accounted for in configuration selection.

Phase Change Materials Results:

PCM configurations require 11.0 lb (5.0 kg) to pass reliably across all months on the US lane, more than three times the mass of the compliant dry ice configuration. The additional thermal stress introduced by the air segment and the higher ambient temperatures along the California last-mile delivery segment drive the substantially higher coolant mass requirement compared to the UK lane.

Gel Packs Results:

Gel pack performance on this lane is notably poor at lower masses. The EPS-IP02.0kg configuration fails in 9% of January profiles and in over 90% of summer profiles. Moving to 6.6 lb (3.0 kg) of gel pack achieves 100% reliability in the EPS container. The performance gap on the US lane reflects the more severe seasonal temperature variation along the route and the additional thermal stress introduced by the air segment. This gap also reinforces a core principle: container and coolant requirements are lane specific. A system that passes qualification on a mild ground route may fail on a route with greater thermal complexity.

Total Cost of Ownership for Each Cooling Medium

Per-shipment packaging cost is the most commonly used metric to evaluate cooling medium options. It is also the least useful in isolation. The relevant metric is cost per compliant shipment at scale, which requires accounting for cooling medium mass, container cost, transportation cost (which scales with total shipment weight), and the cost of product loss from excursions.

UK Lane Cost Analysis

The EPS-DI01.5kg configuration (an EPS container with 3.3 lb (1.5 kg) of dry ice) is the lowest total cost compliant solution at $327,834 annually, which is $28,782 less per year than the gel pack alternative (EPS-IP02.0kg) and $118,638 less than the PCM configuration (EPS-PC03.0kg). The EPS-DI01.0kg configuration, using just 2.2 lb (1.0 kg) of dry ice, appears cheaper on a per-shipment basis, but generates $26,689 in annual product loss, making it the second most expensive option despite its lower packaging cost.

US Lane Cost Analysis

On the US lane, the cost differential widens substantially due to the higher transportation cost per pound on an air route. The dry ice configuration (EPS-DI01.5kg, 3.3 lb / 1.5 kg) is $94,564 less expensive annually than the gel pack alternative (EPS-IP03.0kg, 6.6 lb / 3.0 kg) and $322,714 less expensive than the PCM configuration (EPS-PC05.0kg, 11.0 lb / 5.0 kg). Transportation cost, which scales directly with total shipment weight, is the dominant cost driver on air-connected lanes. Cooling medium mass is therefore not just a packaging decision—it is a freight cost decision.

Across both lanes, the cost data demonstrates a consistent pattern. Dry ice delivers the lowest total annual cost among compliant configurations, and the advantage grows substantially on air-connected routes where transportation cost dominates and coolant mass has a direct multiplier effect on freight spend. The financial case for gel packs and phase change materials weakens further when coolant mass is considered alongside packaging cost. Both alternatives require significantly more coolant to achieve year-round reliability, and that weight penalty compounds across thousands of annual shipments.

The EPS-DI01.0kg result on the UK lane is also worth noting: it is the only configuration that appears cheaper on a per-shipment basis yet delivers the worst total outcome, driven entirely by product loss from summer excursions. This example illustrates the core risk of evaluating cooling medium cost in isolation from thermal performance.

The Carbon Footprint for Each Cooling Medium

Carbon footprint analysis follows the same weight-driven logic as total cost of ownership. The CO₂ equivalent (CO₂e) per shipment is calculated per European Standard EN 16258 and the GLEC Framework, accounting for embodied carbon in packaging materials, transport mode emissions per tonne-km, and route-specific travel distances.

Transportation emissions represent the largest share of total shipment carbon, often by an order of magnitude compared to packaging material embodied carbon. Since transport emissions scale with shipment weight, the total mass of the cooling system is the primary lever for carbon reduction. Container insulation performance plays a secondary, but meaningful role: a better-insulated container requires less coolant mass to achieve the same thermal outcome, which reduces total shipment weight and the associated transport emissions.

On the UK lane, the compliant dry ice configuration (EPS-DI01.5kg, an EPS container with 3.3 lb / 1.5 kg of dry ice) carries 33.3 lb (15.1 kg) total shipment weight. The compliant gel pack configuration (EPS-IP02.0kg, 4.4 lb / 2.0 kg of gel pack) carries 33.9 lb (15.4 kg) and the PCM configuration (EPS-PC03.0kg, 6.6 lb / 3.0 kg of phase change material) carries 37.0 lb (16.8 kg). The differences aggregate meaningfully at 6,000 shipments per year.

The effect is substantially larger on the US air lane. The dry ice configuration (EPS-DI01.5kg, 3.3 lb / 1.5 kg) weighs 31.5 lb (14.3 kg) total; the gel pack configuration (EPS-IP03.0kg, 6.6 lb / 3.0 kg) weighs 34.4 lb (15.6 kg); and the PCM configuration (EPS-PC05.0kg, 11.0 lb / 5.0 kg) weighs 41.0 lb (18.6 kg). At 6,000 annual shipments, the dry ice configuration transports roughly 56,878 lb (25,800 kg) less total payload than the PCM alternative, directly reducing fuel consumption and associated emissions across the air and ground segments of the route.

For organizations with Scope 3 emissions reporting obligations or sustainability commitments tied to logistics operations, cooling medium selection is a carbon lever that is frequently overlooked in favor of higher-profile initiatives when considered alongside container insulation performance.

The carbon data reinforces what the cost analysis shows: the coolant mass required to achieve reliable thermal performance is the dominant variable. On the UK ground lane, the weight difference between the compliant dry ice and PCM configurations is modest at 3.7 lb (1.7 kg) per shipment, but accumulates to over 11,000 lb (5,000 kg) of additional payload annually at 6,000 shipments. On the US air lane, where the PCM configuration requires 11.0 lb (5.0 kg) of coolant against 3.3 lb (1.5 kg) for dry ice, that gap is substantially larger and is amplified by the higher carbon intensity of air freight. For organizations looking to reduce the environmental footprint of their cold chain operations, cooling medium selection is one of the most direct levers available that simultaneously reduces cost.

A Framework for Cooling Medium Selection

Across thermal reliability, total cost of ownership, and carbon footprint, the data from both lanes points consistently in the same direction. The framework that follows translates those findings into a practical decision process that organizations can apply to their own lanes and configurations.

The data across both lanes supports a structured decision framework built around five variables: lane thermal profile, seasonal variation, container insulation performance, payload temperature requirement, and scale.

1) Characterize the lane

Ground routes in temperate climates present lower thermal stress than air routes or lanes crossing high-ambient regions. A lane that passes through summer temperatures consistently exceeding 77°F (25°C) will stress underpowered cooling configurations in ways that qualification testing may not reveal. Historical ambient temperature data, not just worst-case design assumptions, should be the basis for system selection.

2) Identify the seasonal failure window

The most common cold chain failure pattern is a system that performs reliably for eight to ten months per year and fails during peak summer months. If monthly failure rate data is not available, it should be generated before scaling a configuration to operational volumes. A 5% annual failure rate distributed unevenly across months means a significantly higher failure rate in the months that matter most.

3) Account for container and coolant as a system

Container insulation performance directly affects how much of a cooling medium is required to maintain payload temperature. A higher-performing container reduces the thermal load on the coolant and, in some cases, enables a reduction in coolant mass that offsets the higher cost of a better-insulated container. Evaluating containers and coolants independently from one another produces an incomplete picture.

4) Determine the minimum compliant coolant mass for each medium

For each cooling medium under consideration, identify the minimum mass that achieves a 95% or better monthly pass rate across all months on the target lane. Evaluating minimum viable configurations for each option rather than comparing a well-specified solution against an under-specified alternative is the only fair basis for comparison.

5) Calculate total cost of ownership at operating scale

Apply the compliant mass for each medium to a full TCO model: packaging cost, transportation cost based on total shipment weight, and expected product loss based on residual failure rates. At most air-connected and high-volume lanes, the weight efficiency advantages of certain cooling mediums produce cost advantages that are not visible in per-shipment packaging comparisons alone.

Final Takeaway: Dry Ice Outperforms Other Cooling Mediums Across All Criteria

Cooling medium selection is a technical decision with financial and environmental consequences that extend well beyond the cost of the coolant itself. The data from two representative shipping lanes (a 24-hour ground route in the UK and a 24-hour air-connected route across the US) demonstrates that the choice of cooling medium, evaluated against real lane conditions across a full year of historical temperature data, produces meaningfully different outcomes in thermal reliability, total cost of ownership, and carbon footprint.

The analysis also demonstrates that the container and the cooling medium cannot be evaluated in isolation. Insulation performance affects how much coolant is required; coolant mass affects transportation cost and emissions; and the interaction between the two determines whether a configuration is viable across the full seasonal range of a live shipping lane. A system optimized for a single variable (the lowest per-shipment packaging cost, for example) will frequently underperform when evaluated against the full set of metrics that determine operational and commercial viability.

Across the configurations tested, dry ice consistently emerged as the highest-performing cooling medium on all three dimensions evaluated. Dry ice achieved year-round reliability at the lowest coolant mass of any compliant solution at 3.3 lb (1.5 kg) on both lanes. This combination translated directly into the lowest total annual cost and the smallest carbon footprint per shipment.

On the UK ground lane, the compliant dry ice configuration delivered a $327,834 annual total cost against $356,616 for gel pack and $446,472 for phase change material. On the more demanding US air lane, that gap widened to $575,846 for dry ice versus $670,410 for gel pack and $898,560 for phase change material. The performance advantage is not marginal, and it compounds at scale. For direct-to-consumer food shippers evaluating cooling medium options across high-volume lanes, the data makes a clear case for dry ice as the solution that best aligns operational reliability with cost efficiency and sustainability objectives.

The margin between a configuration that fails 64% of July shipments and one that passes every profile year-round can be as small as 1.1 lb (0.5 kg) of coolant. That precision matters, and it is only achievable through lane-specific data and thermodynamic modeling against real historical conditions. The analysis presented here offers both a methodology and a finding: dry ice, correctly specified for the lane, is the cooling medium best positioned to deliver on reliability, cost, and carbon across the demands of modern cold chain logistics.

Want to learn more about using dry ice for food shipping? Check out our resources on how best to utilize dry ice for delivering product to specification!