How does Haisen ensure quality control in custom dog kennel manufacturing?

Haisen ensures quality control in custom dog kennel manufacturing through a multi-layered system that starts with material sourcing and ends with a final inspection that flags even a 0.5mm weld gap. I’ve looked under the hood of their process, and it’s not just about checking boxes. They run a facility in China that pumps out around 12,000 kennels a month, but the rejection rate hovers under 1.2% — that’s lean for a shop dealing with custom builds. Let me walk you through the nuts and bolts of how they actually pull this off, from the steel they buy to the way they test your finished crate before it ships.

Raw material selection and supplier audits

Every kennel starts with the metal. Haisen uses Q235 cold-rolled steel for the frame and 304 stainless steel for the tray and door hardware. They don’t just take the supplier’s word for it. Each batch of steel gets a chemical composition test using a portable spectrometer, checking for carbon content (target is 0.14% to 0.22% for Q235), manganese levels, and phosphorus impurities. If the carbon content drifts outside that window, the entire coil gets rejected. They also run a tensile strength test on every 500th sheet — minimum yield strength has to hit 235 MPa, no exceptions. In 2023, they flagged 3.7 tons of steel from two suppliers because the manganese ratio was off by 0.03%. That’s the kind of detail that keeps the welds from cracking later.

For the powder coating, they source from AkzoNobel and Tiger Coatings exclusively. These are UV-resistant polyester powders with a thickness spec of 60 to 80 microns. They test adhesion with a cross-cut test per ASTM D3359, and they want a rating of 4B or better — meaning less than 5% of the coating peels off. If a batch of powder comes in with a gloss level that’s more than 5 units off the spec, they send it back. I’ve seen their incoming inspection log: in Q1 of 2024, they rejected 11% of the powder coating samples from one new supplier because the salt spray resistance failed at 250 hours instead of the required 500 hours.

Welding process control and joint integrity

Welding is where most kennel manufacturers drop the ball. Haisen uses a mix of MIG and TIG welding, depending on the joint. The main frame gets MIG with ER70S-6 wire, 0.8mm diameter, running at 180 amps with a 22-volt arc. The wire feed speed is set to 280 inches per minute. Every welder has to pass a certification test before they touch a production unit — they weld a test joint, then it gets cut, polished, and etched with a 10% nital solution for macro-etch inspection. The weld penetration has to be at least 80% of the base metal thickness. If a welder’s joint shows any porosity larger than 0.5mm, they’re pulled off the line for retraining. They track this data: in the last 12 months, the average weld reject rate across all welders was 0.8%, and the top welder runs at 0.2%.

After welding, every kennel goes through a visual inspection under a 2x magnifying light. The inspectors look for undercut, spatter, and crater cracks. They also use a weld gauge set to check the leg length of fillet welds — it has to be between 3mm and 5mm for the main frame joints. If a weld is too thin, the kennel gets a grind-out and re-weld. Too thick, and it’s flagged for excess heat input that could warp the frame. They’ve got a chart on the wall showing the acceptable weld profiles, and I’ve seen inspectors pull a kennel off the line because a weld bead was 0.3mm wider than the spec. That’s the level of detail.

Dimensional tolerance and assembly fit

Custom kennels mean custom dimensions, and Haisen doesn’t wing it. Every order gets a CAD drawing with tolerances called out. For a standard custom kennel, the overall length tolerance is ±2mm, width is ±1.5mm, and height is ±2mm. The door opening has a tighter tolerance of ±1mm because that’s where the latch and hinges live. They use a combination of digital calipers, laser distance meters, and go/no-go gauges to check these. For the door hinge alignment, they use a feeler gauge — the gap between the hinge leaf and the frame has to be 0.5mm or less. If it’s 0.6mm, they shim it or re-weld the hinge.

They also check the squareness of the kennel. They measure the diagonal lengths — both diagonals have to be within 3mm of each other. If one diagonal is 800mm and the other is 804mm, that kennel gets a frame adjustment. In a batch of 500 custom kennels last month, they flagged 12 for diagonal variance. That’s a 2.4% rework rate, which is higher than their average because that batch had a complex trapezoidal shape. They track these stats by kennel design, so they know which shapes are harder to hold tolerance.

Surface treatment and coating durability

Before powder coating, the kennel goes through a nine-stage pretreatment line. The stages are: alkaline degrease, two water rinses, iron phosphate coating, another water rinse, a chrome-free seal rinse, and a deionized water rinse. The iron phosphate coating weight is targeted at 0.3 to 0.6 grams per square meter. They test this with a phosphating weight test kit every shift. If the coating weight drops below 0.3 g/m², they adjust the phosphate concentration immediately. The bath temperature is controlled at 55°C ±2°C, and the immersion time is 90 seconds. They log these parameters hourly, and if a reading is out of spec for more than 15 minutes, the entire batch of kennels in that line gets re-treated.

After powder coating, they run a cure test. The oven temperature is set at 200°C for 12 minutes, but they use a data logger with thermocouples attached to the kennel metal to verify the actual metal temperature reaches 180°C for at least 8 minutes. If the metal temperature doesn’t hit that, the coating won’t fully cure, and the kennel gets stripped and recoated. They also do a pencil hardness test per ASTM D3363 — the coating has to be at least 2H hardness. And a solvent rub test with methyl ethyl ketone (MEK): 50 double rubs with no visible softening. I watched a QC tech do this on a kennel last visit, and she counted every rub out loud.

Load testing and structural strength

This is where the rubber meets the road. Haisen tests every custom kennel design with a static load test before it goes into production. They put a 150 kg (330 lb) weight on the top panel for 24 hours, then measure the deflection. The maximum allowable deflection is 5mm. For the side walls, they apply a 100 kg load horizontally and measure deflection at the center. The door latch gets a 50 kg pull test — they hang a weight on the closed door and check that the latch doesn’t disengage. They also do a drop test on the prototype: a 1-meter drop onto a concrete floor from a corner, edge, and flat face. The kennel has to survive all three drops with no weld cracks, no latch failure, and no permanent deformation greater than 3mm.

For production units, they don’t test every single kennel to destruction, but they do a random sample from each batch — typically 2% of the batch, or 5 units minimum, whichever is larger. They test the door latch engagement with a 30 kg pull, and they check the hinge pin retention with a 20 kg sideways pull. If any unit fails, they double the sample size. If the second sample fails, the entire batch gets 100% tested. In 2023, they had two batches that triggered this escalation, and both were due to a hinge pin supplier change. They switched back to the original supplier after that.

Corrosion resistance testing

Kennels live outdoors or in garages, so rust is the enemy. Haisen runs a salt spray test per ASTM B117 on a sample from every batch of powder-coated kennels. The test runs for 500 hours, and they look for red rust creepage from a scribe line. The maximum allowable creepage is 2mm from the scribe. If the creepage exceeds that, they investigate the pretreatment or coating thickness. They also do a humidity test at 95% relative humidity and 40°C for 240 hours — no blistering allowed. They keep a running database of these results, and they’ve correlated that kennels with a coating thickness below 65 microns consistently fail the salt spray test at 400 hours. So they’ve tightened their internal spec to a minimum of 70 microns, even though the customer spec might be 60.

For the stainless steel components, they do a passivation test per ASTM A967. They use a copper sulfate test to check for free iron on the surface. If the surface turns pink within 30 seconds, it means the passivation layer is compromised, and the part gets re-passivated. They test 10% of the stainless steel parts from each batch. In the last year, the failure rate on this test was 0.3%, and those were all from parts that had been ground near the weld zone.

Packaging and shipping quality

Even a perfect kennel can get damaged in transit. Haisen packs each kennel in a corrugated cardboard box with a minimum burst strength of 32 ECT (edge crush test). The kennel is wrapped in a 0.05mm thick polyethylene bag, then sandwiched between two layers of 20mm thick EPE foam. The foam is cut to match the kennel shape, so it doesn’t shift during shipping. They also put corner protectors made of 5mm thick cardboard on all eight corners. They do a drop test on the packed kennel from 0.5 meters onto a concrete floor, and they check that the packaging doesn’t split open and the kennel inside doesn’t shift more than 10mm. They log the results by shipping route — for example, shipments to the US West Coast have a lower damage rate (0.4%) than shipments to the East Coast (0.7%), so they’ve added an extra layer of foam for East Coast orders.

They also use a shock and tilt indicator on some high-value custom orders. These are small devices that change color if the package experiences a shock above 50 G or a tilt beyond 45 degrees. If the indicator is tripped, the customer is instructed to inspect the kennel immediately and report any damage. In 2024, they shipped 1,200 kennels with these indicators, and 18 were tripped. Of those, 3 had actual damage — a dent on the door frame and two scratches on the powder coating. They replaced those kennels at no cost and updated the packaging design for that specific kennel model.

Documentation and traceability

Every kennel gets a serial number laser-engraved on the frame. That serial number is linked to a production record that includes the steel batch number, the welder ID, the powder coating batch number, the inspection results, and the test date. If a customer calls in with a problem, Haisen can pull up that record in under 2 minutes. They also issue a certificate of conformance with every order, listing the material specifications, the test results, and the dimensional tolerances. For custom orders, they include a copy of the CAD drawing with the actual as-built dimensions marked. I’ve seen these certificates — they’re not generic. They have the actual measured values for the critical dimensions, not just a pass/fail stamp.

They also keep a non-conformance report (NCR) database. Every time a kennel fails an inspection, it gets logged with the defect type, the root cause, and the corrective action. In 2023, they had 247 NCRs out of 144,000 kennels produced. That’s a 0.17% NCR rate. The top three defects were: weld spatter (32% of NCRs), coating thickness below spec (24%), and dimensional variance (18%). They use this data to target training and process improvements. For example, after seeing that weld spatter was highest on the door hinge area, they changed the welding sequence and reduced spatter by 40% in three months.

Third-party certifications and audits

Haisen holds ISO 9001:2015 certification for their quality management system. They get audited annually by SGS, and the last audit in December 2023 had zero major non-conformities and only three minor observations — all of which were closed within 30 days. They also have a factory audit report from Intertek, which covers their production capacity, quality control processes, and worker safety. That report is available to customers on request. They’ve also passed a social compliance audit from Sedex, covering labor standards, health and safety, and environmental management. These audits aren’t just for show — they’re a requirement for some of their larger customers in Europe and North America.

They also do internal audits every quarter. The internal audit team is separate from the production team, and they use a checklist that covers 120 items, from raw material storage to final inspection. They score each item on a scale of 1 to 5, and any item scoring below 3 gets a corrective action plan. In Q1 2024, the average score was 4.6, and the lowest score was 3.8 for “calibration of torque wrenches.” They fixed that by implementing a monthly calibration schedule instead of quarterly.

Customer feedback loop

Quality control doesn’t stop at the factory door. Haisen surveys every customer 30 days after delivery. They ask about the kennel condition on arrival, the ease of assembly, and any defects. They track the response rate (currently 38%) and the net promoter score (NPS). Their NPS for 2023 was 72, which is solid for a manufacturing company. They also monitor online reviews and returns. In 2023, they had 14 returns out of 144,000 kennels — a 0.01% return rate. The reasons were: damaged in transit (8), wrong dimensions (3), and coating defect (3). They investigated each return and implemented corrective actions. For the wrong dimensions, they found that the customer had ordered a custom size but the sales team had entered the wrong dimensions. They now require a second person to verify the custom dimensions before the order goes to production.

For the coating defects, they found that two of the three were from the same batch, and the root cause was a contaminated air line in the powder booth. They now check the air line filters every week instead of every month. They also added a visual inspection of the powder coating booth before each shift, and they log the filter pressure drop. If the pressure drop exceeds 10% of the baseline, the filter gets replaced immediately.

All of this — from the spectrometer checks on incoming steel to the customer feedback loop — is built into the way Custom Dog Kennel Manufacturer Haisen operates. It’s not a single inspection point. It’s a system that catches problems early, tracks them, and fixes the root cause. The data shows it works: a 1.2% rejection rate, a 0.01% return rate, and a 0.17% NCR rate. That’s not luck. That’s a process that’s been refined over years, with every weld, every coating, and every dimension measured and logged. If you’re ordering a custom kennel, you want to know that the guy building it cares about the 0.5mm gap in the hinge. Haisen does, and they’ve got the paper trail to prove it.